Upgrading the Level II Protocol for Physiological Modelling
of Cause-effect Relationships: A Pilot Study
Final Report for EC project No. 98.60.UK.003.0 – a pilot and demonstration project of EEC regulation 3528/86
and EC 1091/94 on the protection of the Community’s forests and against atmospheric pollution.
Broadmeadow M.S.J., Pitman R.M., Jackson S.B., Randle T.J. and Durrant D.W.H.
Forestry Commission Research Agency
Alice Holt Lodge
Wrecclesham
Farnham
Surrey, GU10 4LH
UK
Executive summary
EU Council Regulation 3528/86 for the protection of the forests of the European Community
requires that a monitoring programme is carried out to ensure the protection of the
Community’s forests. To date, this monitoring programme has concentrated on the impacts of
acid deposition, together with a number of projects which have focussed on developing the
intensive forest health monitoring programme to encompass a wider range of both natural and
anthropogenic drivers of forest health within the EU. Process modelling at the physiological
level would contribute significantly to this more holistic approach and could be made possible
through a development of the monitoring protocol defined within Council Regulation
1091/94. This modelling strategy would further the programme’s ability to identify and
predict the effects of ozone pollution, rising atmospheric carbon dioxide concentrations,
global climate change, and interactions between these influences on vitality and yield of the
community’s forests, including the potential for carbon sequestration. Furthermore, it would
enable more precise estimates of pollutant load to forest ecosystems to be made, and develop
the capacity to derive cause-effect relationships.
The analysis of interactions between site factors and forest health and growth at the
physiological level requires an intensive modelling approach, which is at one extreme of the
modelling spectrum. Other models of forest growth are available which operate at different
temporal and spatial scales, and require specific datasets which reflect these scales. The work
described here, and the recommendations made from it should therefore be placed in the
context of the intensive modelling of forest function, an approach that will complement the
data analysis methodologies that are already in place.
The objective of this report is to demonstrate how the Level II monitoring protocol can be
upgraded to provide sufficient information of sufficient quality to enable intensive modelling
of forest processes to be undertaken, with an emphasis on the application of modelling to
derive cause-effect relationships relating to climate and ambient air quality, including a
predictive capability.
Two of the ten Intensive Forest Health monitoring plots (Level II) in the UK have been
instrumented to provide additional biotic and abiotic information to enable process driven
modelling of ecosystem function to be undertaken. Both sites are plantations of oak, but the
climate, soil type, topography and canopy structure differ greatly. The existing protocol for
routine site visits (for deposition sampling and analysis) has also been augmented to provide
data for model parameterisation and validation. Additional instrumentation has included a
within-plot automatic weather station, soil moisture probes, automatic girth bands and
radiation sensors mounted above and below the canopy to provide an indication of light
transmission and thus canopy development. Litter collection and foliar sampling of different
positions within the canopy have been carried out on a monthly basis, whilst routine
fortnightly visits have included downloading of the weather station data and manual
measurement of soil moisture.
Annual carbon and water fluxes have been simulated using the process model ForestFlux
(formerly GROMIT), which is an assemblage of widely accepted process descriptions
including photosynthesis, evapo-transpiration, respiration, light interception and soil water
balance operating at an hourly timestep. The parameters and variables derived during the
course of this project have been used as inputs for ‘baseline simulations’, which have been
compared with a range of other simulations in which a selection of key input parameters and
variables have been altered to provide a sensitivity analysis of their impact on modelled
ecosystem function.
2
The combination of autumn litter collection and analysis and continuous recording of canopy
light transmission together with spot measurements of light transmission throughout the
mensuration plot have provided a record of canopy development. However, the use of
automatic girth bands as indicators of budburst was unsuccessful as a result of other
environmental factors, particularly soil moisture and bark hydration, interfering with the
signal of stem girth changes at the time of budburst.
Monthly sampling of foliage was successfully carried out using the meteorological mast to
access the canopy. Nitrogen analysis of these samples permitted the derivation of parameters
describing photosynthesis and respiration. After an initial site specific calibration period, it is
suggested that specific leaf area (area per unit dry weight of leaf) could be used as a surrogate
for foliar nitrogen content, given the strong relationship observed at both sites. The analysis of
foliage from throughout the canopy, rather than a single point from the upper canopy, had a
particularly large impact on model outputs, and in combination with a comprehensive analysis
of the distribution of specific leaf area at the time of autumn leaf fall, a full description of
canopy photosynthetic function can be derived.
The continuous measurement of soil moisture provided partial validation for the water
balance sub-model, or alternatively, is available as a direct model input. The markedly
different water release characteristics of organic top soil and mineral sub-soils at the two sites
studied here required that soil moisture was monitored in both.
Climatic input variables were measured at three different meteorological stations, which
allowed a comparison of simulation results using the three datasets: (a) daily climatological;
(b) hourly automatic adjacent to (a); (c) within-plot automatic. This sensitivity analysis
indicated that hourly readings of both rainfall and solar radiation were essential, as the precise
distribution of both impinges to a large extent on annual water and carbon balance,
respectively. Soil temperature was considerably lower beneath the canopy compared with the
open climatological station site, which significantly reduced modelled soil/root carbon
emissions. Other variables had little impact on model outputs, although the difference in
minimum air temperature between the climatological station and the forest plot was
considerable at times. Although this difference in minimum air temperature did not affect
carbon balance on an annual basis, analysis of poor forest condition and yield loss as a result
of frost damage would benefit greatly from air temperature data measured at the forest plot.
The sensitivity analysis of model parameters revealed that canopy development, light
extinction coefficients and parameters describing photosynthetic capacity had the greatest
impact on annual carbon balance, all of which can be derived using the approach adopted in
this demonstration project.
The modelling approach described in this report indicates that the measures adopted here
provide sufficient data to enable process evaluation of forest function to be undertaken. These
measures do not provide a comprehensive suite of all model parameters that may be required,
but will allow an analysis of the relative impact of the many drivers of forest condition that
act at the physiological level. Furthermore, to gain fully from the resource investment, where
sites are upgraded, this should coincide with the measurement of ambient air quality and other
optional variables within the framework of the monitoring programme. Careful consideration
should also be given within a Europe-wide context, to both the choice of species and
distribution of plots that are represented in any upgraded network.
3
The following recommendation is given as a minimum for upgrading broadleaf forest plots to
enable process modelling at a sub-daily scale:
• the installation of a within-plot automatic weather station measuring the following
variables (as a minimum):
soil temperature
air temperature
soil moisture (organic and mineral horizons)
solar radiation above and below the canopy
• the installation of litter collectors including an annual analysis of specific leaf area
• foliar sampling on an annual basis, including an assessment of variation within the
canopy; ideally this should be carried out on a monthly basis.
4
Table of Contents
EXECUTIVE SUMMARY ......................................................................................................2
TABLE OF CONTENTS .........................................................................................................5
1 INTRODUCTION .................................................................................................................6
2 METHODOLOGY ................................................................................................................9
2.1 SITE DESCRIPTIONS ............................................................................................................9
2.1.1 Grizedale ...................................................................................................................9
2.1.2 Alice Holt.................................................................................................................10
2.2 EQUIPMENT .....................................................................................................................12
2.2.1 Meteorology.............................................................................................................12
2.2.2 Girth.........................................................................................................................12
2.2.3 Soil Moisture............................................................................................................12
2.2.4 Canopy Development and Leaf Area.......................................................................13
2.3 PROTOCOLS .....................................................................................................................13
2.3.1 Automatic data collection and weather station maintenance..................................13
2.3.2 Phenological development.......................................................................................14
2.3.3 Foliar chemistry ......................................................................................................15
2.3.4 Litter collection........................................................................................................15
2.3.5 Derivation of light extinction coefficient (Kext) and leaf area index (LAI)..............16
2.3.6 Soil moisture ............................................................................................................16
3 RESULTS .............................................................................................................................18
3.1 METEOROLOGY ...............................................................................................................18
3.2 CANOPY DEVELOPMENT ..................................................................................................23
3.3 GIRTH INCREMENT AS AN INDICATOR OF BUDBURST ........................................................29
3.4 FOLIAR CHEMISTRY .........................................................................................................32
3.5 SOIL MOISTURE ................................................................................................................35
3.6 MODELLING .....................................................................................................................38
3.6.1 Approach .................................................................................................................38
3.6.2 Baseline simulations ................................................................................................38
3.6.3 Respiratory components ..........................................................................................39
3.6.4 Sensitivity analysis of meteorological data .............................................................42
4 CONCLUSIONS..................................................................................................................48
5 RECOMMENDATIONS ....................................................................................................49
ACKNOWLEDGEMENTS ...................................................................................................50
REFERENCES .......................................................................................................................51
APPENDICES ..........................................................................................................................54
I Model description and Validation of ForestFlux (GROMIT) ........................................54
II Reporting forms ............................................................................................................70
III Pictorial representation of standard flushing stages ..................................................74
IV Glossary of abbreviations and definition of symbols...................................................75
5
1 Introduction
The Level I network of forest health monitoring plots was set up against a background of acid
deposition and dramatic forest defoliation. It has provided a wealth of data allowing statistical
analyses of the impact of a series of drivers relating to meteorology, deposition and other site
factors (UN/ECE and EC, 1997). More detailed, site specific parameters are measured in the
sub-set of sample plots included in the Intensive Forest Health Monitoring Network (Level
II), although these more detailed measurements are largely orientated towards acid deposition.
As a result of policies adopted by individual member states of the European Union, there has
been a significant reduction in acid deposition, largely due to emission control measures
implemented in power stations and other heavy industry resulting in rapidly falling sulphur
dioxide concentrations. Over this same period, other environmental issues have become more
prevalent, both in the media, and also in the decision making processes of the European
Union and member states.
Anthropogenic carbon dioxide emissions from the industrial revolution to the present day
have resulted in ambient CO2 levels in the atmosphere rising from approximately 270 ppm to
370 ppm. These emissions will continue, and the concentration is expected to rise to between
600 and 700 ppm by the end of this century (IPCC, 1995). There is now convincing evidence
that the rise in atmospheric CO2 concentrations have resulted in a gradual warming of the
climate, culminating in five of the warmest summers on record (global average) having
occurred during the 1990s. This increase in global temperatures will be accompanied by
changes to other climatic variables including rainfall distribution and patterns, cloud cover,
potential evapo-transpiration and the occurrence of storms. As a consequence, it is essential
that any forest health monitoring programme has the capacity to disagregate the effects of
climate change from those of air pollution and acid deposition, both to allow in-depth
analyses of existing data, and also to enable predictions of future changes in the condition of
European forests.
Issues associated with ozone pollution are increasingly being considered to have an important
bearing on forest condition, particularly in southern Europe (Sanz and Millan, 1998). Ozone is
a secondary pollutant, and is produced by photochemical reactions in the presence of other
pollutants such as volatile organic compounds (VOCs) and nitrogen dioxide. As a result (of
being a secondary pollutant), it is difficult to predict its current distribution, and also, to
project how concentrations will change in the future. Furthermore, its phytotoxic action is
mediated through the stomata, and as a consequence, concentration related damage indices are
not suitable under some conditions, particularly under conditions of drought which often
coincide with ozone episodes. Detailed analyses of the effects of ozone pollution on European
condition thus require the calculation of physiologically effective ozone doses, rather than the
monitoring of cumulative exposure (Broadmeadow, 1998; Broadmeadow et al., 1999;
UN/ECE, 1996). In the same way, total nitrogen inputs to any ecosystem require that uptake
mediated by the stomata of the highly reactant species NH3, and to a lesser extent NO2 is
calculated. In both cases, estimates of deposition by the measurement of throughfall do not
account for this contribution to dry deposition, which certainly in the case of NH3, is
important.
A number of process models of forest function and growth have been developed over recent
years. These operate at a number of different spatial and temporal scales, and represent
different processes at varying levels of complexity. Models such as BIOMASS (McMurtrie et
al., 1990), G’DAY (Comins and McMurtrie, 1993), and FINNFOR (Kellomäki et al., 1993;
Väisänen et al., 1994) operate at long temporal scales and have been developed for long term
modelling, including an assessment of decadel and centural changes in soil nutrient balance.
6
Other models such as HYBRID (Friend et al., 1997) have been developed for regional scale
modelling in the long term. At the other extreme, canopy exchange models such as PGEN
(Friend, 1995) and MAESTRO (Wang and Jarvis, 1990; Jarvis, 1993) represent canopy
structure, radiation balance and gas exchange in great detail over short timeframes, but do not
incorporate long-term growth functions. There are also a small group of models (such as
GROMIT: Randle and Ludlow, 1998; Broadmeadow et al., 1999) which operate at a
hourly/daily timestep, but are integrated with a growth module to enable commercial rotation
length simulations to be performed. Reviews of modelling approaches are available (eg
Sonntag, 1997). It is essential that the spatial and temporal resolution of the model chosen is
appropriate to the analysis being undertaken, since the resource input that is required for
modelling at the range of scales given above varies greatly.
Process models operating at an hourly/daily timestep are intensive in terms of meteorological
data input. A number of site specific parameters are also required, in addition to species
specific parameters that in many cases can be obtained from the scientific literature. In order
for representative model simulations to be carried out for the range of intensive monitoring
sites set up under the UN ECE/ICP Forests forest health network, a number of decisions have
to be made requiring data capture and input to a suitable model.
•
•
•
•
•
•
•
•
•
Will process modelling provide further cause-effect information?
What is the spatial and temporal resolution that best fits the anticipated needs of the
Programme?
Is there an associated weather station?
Are the conditions measured at the meteorological representative of the forest plot?
Which additional plot specific measurements are required as inputs to enable
process modelling of cause effect relationships?
Are additional measurements practicable given available resources?
How frequently should the additional measurements be made?
How sensitive are the model outputs to the additional measurements?
Can validation exercises be undertaken?
The objectives of this pilot study are thus, firstly, to demonstrate how a minimum number of
additional measurements can be made within the existing framework of routine site visits
under the Level II protocol, to provide sufficient site specific variables and input parameters
to enable a process based assessment of forest function. The second role is to indicate the
relative sensitivity of model outputs to the additional measurements carried out under this
pilot study. Finally, recommendations will be made as to which measurements should be
included within the level II protocol, and which further measurements should be considered to
further enhance the value of a process modelling approach. It should be stressed that this
report focuses on process modelling at a sub-daily time interval and is intended to be
applicable specifically to the impact of ozone pollution and global climate change, including a
predictive capability. Recommendations may therefore not be appropriate to other modelling
approaches, and should be set against a strategy for the analysis of the Level II dataset which
includes modelling at a variety of spatial and temporal scales.
7
Before the detailed recommendations that are made here are implemented, two hypotheses
which are outside the scope of this report should be explored to demonstrate a clear benefit of
the process modelling approach and associated site specific measurements; without an
unambiguous benefit, and potential for improved derivation of cause-effect relationships, the
additional resources required for this approach cannot not be justified
1. Process modelling of forest function and growth provides further insight into causeeffect relationships compared to multivariate statistical analysis of existing core
measurements under the Intensive Monitoring Protocol.
2. Plot specific measurements improve analyses of forest function, over interpolated
datasets; this is specific to some areas of the programme, including the impact of ozone
pollution and climate change and rising atmospheric CO2 concentrations.
3. Process modelling can be used as a predictive tool to identify the anticipated effects of
global environmental change on the Communities forests.
8
2 Methodology
2.1 Site descriptions
Two of the three oak (Quercus spp.) plots within the UK Level II network (see Fig. 1) were
selected representing contrasting soil, deposition and meteorological conditions. Site 517
(Lakes or Grizedale) was upgraded from a standard intensive monitoring plot, whilst site 512
(Alice Holt) had some additional instrumentation at the time of installation, and some model
parameterisation had been carried out at this site during the previous two years. During the
course of the study, an Edisol eddy correlation flux measurement system (see Aubinet et al.,
1999) was installed to improve model parameterisation and enable process validation.
Fig. 1. Location of the ten Intensive Forest Health Monitoring Plots in the UK. The two sites used in this study
are Grizedale (Lakes) and Alice Holt.
2.1.1 Grizedale
Fig. 2 Grizedale (Level II) Intensive Forest Health Monitoring Plot.
9
An oak plantation (54o19’N; 3o2’W) established in 1920 at an altitude of 120 m (see Fig. 2).
The plot is within a mixed species forest with rolling/mountainous topography. Mean top
height is 18.4 m, DBH 30.2 cm, and stocking density 310 trees per hectare (under-stocked at
present) resulting in a basal area of 20 m2 ha-1. The soil is a brown podzol (Pyatt, 1982)with
an effective rooting depth of 50 cm. Mean annual precipitation is 1800 mm. Understorey
vegetation cover is approximately 50%, and is dominated by grasses, bilberry, bracken and
mosses. A long-term weather station is located within 2 km. General yield class (GYC) is 4
(see Edwards and Christie, 1981), and total nitrogen deposition approximately 19 kg ha-1 a-1.
Site layout is shown in Fig. 3, including the position of the additional collectors and
instrumentation installed for the purposes of this pilot study.
Fig. 3 Typical layout of UK Level II Intensive Forest Health Monitoring Plot (Grizedale).
2.1.2 Alice Holt
A relatively homogeneous and mono-specific forest block planted in the 1930s and covering
an area of approximately 70 ha (see Fig. 4). The altitude is 80 m and the plot is situated within
Alice Holt forest in south-east England (51o10’N; 0o51’ W). Other species (mostly ash) make
up 10% of the tree cover and the understorey is dominated by hazel (Corylus avellana),
hawthorn (Crataegus monogyna), Rubus spp. and various grass and herbaceous species. The
soil is a surface-water gley (Pyatt, 1982) with a depth of 80 cm to the C horizon of the
cretaceous clay. The pH is 4.6 and 4.8 in the organic and mineral horizons respectively. Top
height and DBH were 19.3 m and 25.9 cm respectively in 1995 at a density of 606 trees per
hectare resulting in a basal area of 22 m2 ha-1; general yield class is 6 and the site was last
thinned in 1995 (and 1991). A UK Environmental Change Network plot has also been set up
within the forest block (ECN, 1996). Daily meteorological data are available from 1955
(within 5 km of the stand), and an automatic weather station was installed in 1994. Total
nitrogen deposition was 9.1 and 7.4 kg ha-1 in 1996 and 1997 (calculated after Ulrich, 1983),
respectively, and a continuous pollution record (hourly concentrations of SO2, NOx, O3) is
available from 1987 (and NH3 from 1996). Mean annual rainfall is 780 mm, and mean annual
temperature 10.6oC. The meteorology mast and instrumentation are located 200 m from the
Intensive Monitoring Plot, at the centre of the plantation to provide a 500 m fetch for the
micrometeorology measurements.
10
(a)
(b)
(c)
Fig. 4. (a) Alice Holt (Level II) Intensive Forest Health Monitoring plot (b) Alice Holt meteorology and flux
measurement mast (c) view of Alice Holt meteorology and flux measurement mast from above the canopy.
11
2.2 Equipment
2.2.1 Meteorology
Both sites have associated long term UK Meteorological Office climatological stations. These
are sited 2 km and 4 km distant from the forest plot at Grizedale and Alice Holt, respectively.
Process model inputs provided by these climatological stations are maximum and minimum
daily temperature, run of wind, 9 am vapour pressure (from wet and dry bulb temperature),
soil temperature and rainfall. Sunshine duration is recorded at Alice Holt, but not Grizedale.
Alice Holt is also in the process of transferring to an automatic weather station (AWS) and for
the duration of this project, both automatic (hourly) and manual (daily) measurements were
made. The automatic weather station is operated according to the protocol laid down by the
UK Environmental Change Network (ECN, 1996). Both sites have also been instrumented
with within plot weather stations and meteorological masts recording the following variables:
• 1Rainfall (0.2 mm tipping bucket rain gauge: Didcot Instruments, Oxford, UK)
• 2Above canopy wet and dry bulb temperature (non-aspirated psychrometer with platinum
resistance elements (PT385: ELE International, Hemel Hempstead, UK).
• Soil temperature at 30 cm (Grade A PT385: RS Supplies, Corby, UK).
• Above and below canopy global radiation (tube solarimeters: ELE International, Hemel
Hempstead, UK)
• Above canopy net radiation (net radiometer: ELE International)
• Wind speed (cup anemometer: Vaisala, Sweden)
• Wind direction (wind vane: Vaisala, Sweden)
• Soil moisture at 30 cm (Theta probe ML1: Delta T Devices, Cambridge, UK)
1
Rainfall was measured in a clearing adjacent to the plot at Grizedale; at Alice Holt, no clearing was available
adjacent to the plot, and hourly rainfall data are those for the automatic weather station adjacent to the long-term
climatological station (4 km from the forest plot).
2
The psychrometer reservoir was refilled with distilled water every two weeks during routine site visits, using a
horticultural spray to pump the water to the top of the meteorological mast.
At both sites, variables were measured every 10 seconds and recorded as 30 minute averages
(or totals in the case of rainfall) using 10/30 channel data loggers (DT500: DataTaker,
Letchworth, UK) powered by 10 W solar panels with backup provided by a 72 Ah lead acid
accumulator. The weatherproof enclosures were mounted at 15 m on the meteorological mast
to reduce the potential for vandalism. In addition, the weather station data logger also
collected automatic girth band data (section 2.2.2).
The masts were constructed from 50 mm tubular aluminium to form 450 mm square sections
either 2 m or 3 m in length, and mounted on railway sleepers bedded in sand and supported by
one set of 3 mm steel guy wires for every 8 m height of mast. Screw-in ground anchors (0.75
m long) secured the guy wires. The structure also provided canopy access for foliar sampling
and routine weather station maintenance.
2.2.2 Girth
Automatic girth bands (D6: UMS GmbH, Munich, Germany) were installed on four trees in
each plot. Stainless invar steel wire ran over teflon netting (to reduce surface resistance) and
operated a four wire Wheatstone full-bridge strain gauge. A stabilised 5V power supply was
provided by the data-logger.
2.2.3 Soil Moisture
In addition to the theta probe (ML1) associated with each within plot weather station, two
TDR probes (Earth Science Systems, Hemel Hempstead, UK) were permanently installed at
each plot. These 120 cm long, five section probes provided manual measurements of soil
moisture on each routine site visit at depths of 0-15 cm, 15-30 cm, 30-60 cm, 60-90 cm and
12
90-120 cm. At Grizedale, bedrock was 60-90 cm from the surface at the deepest point and as
a consequence, only 2 or three sections were within the soil. A 75 cm probe was used to
assess the spatial variation in soil moisture corresponding to the thirty locations at which soil
descriptions were recorded.
2.2.4 Canopy Development and Leaf Area
Canopy development was derived from light transmission measurements made using the two
tube solarimeters on the meteorological mast. Leaf area index was calculated according to
Goudriaan (1988) with the effective light extinction (Kext) coefficient obtained from inverting
the same model using measured LAI from litter collection and spot measurements of light
transmission (τ) in the Intensive Monitoring Plot during 1998 and 1999. Spot measurements
of LAI above the litter traps were only made under overcast conditions resulting in the diffuse
fraction assuming a value of 1.0, and as a consequence, the assessment of leaf angle
distribution was not required. Further details of canopy structure and leaf distribution for
Alice Holt are given in Kull et al. (1999). As a consequence of the methodology adopted for
the continuous estimation of LAI, data analysis of the light transmission record was restricted
to 1000-1400 h (solar time), and when global radiation was less than 200 mW m-2 (ie beam
radiation approximated zero during the leafed period).
2.3 Protocols
2.3.1 Automatic data collection and weather station maintenance
Data were downloaded during routine fortnightly site visits. A batch file enabled a single
command from the command prompt to activate the software, download the data to a lap-top
hard drive, delete the data from the logger memory, and copy the data to a removable disc for
forwarding to the project manager. In addition to downloading data, routine site visits
included a visual inspection of wiring. Data were received the following day by the project
manager allowing data integrity to be evaluated. Any necessary action was either carried out
on the next scheduled site visit, or by an unscheduled site visit where major technical
problems were evident. Routine maintenance was carried out on a six monthly basis,
including the following:
•
•
•
•
•
•
•
clean and check functionality of rain gauge
replace net radiometer domes
replace wick in psychrometer
check wet and dry bulb readings within 0.2o with no wick present
check operation of wind vane through 360o
check state of anemometer bearings
flush tube solarimeters through with dry air to remove moisture and prevent
condensation; remove any algal growth from solarimeters
For modelling purposes (see section 3.5), missing data (arising from equipment failure or
malfunction) at Alice Holt have been substituted by AWS data from other local stations
operated by Forest Research, after correction for site differences using linear regression (see
section 3.1). No alternative AWS was available at Grizedale, and as a consequence, missing
data was substituted by hourly data generated from daily climatological data and correcting
for site differences using linear regression. No radiation input (sunshine duration) is recorded
at the climatological station at Grizedale – an estimate of global radiation was synthesised
from daylength and latitude with cloud cover assumed on raindays according to Evans (1995)
after Gates (1980) and Nikolov and Zeller (1992). The estimated radiation sum for 1999 was
scaled to the measured radiation input from the canopy AWS (except days when the AWS
radiation record was incomplete), and this scaling factor was used throughout the year to
account for cloud cover on days receiving no rainfall.
13
2.3.2 Phenological development
The object of developing an accurate method of recording and monitoring spring budbreak,
was to provide information on when the juvenile foliage, unprotected by bud scales, is
susceptible to frost damage. Data collected using this or a similar protocol at forest stand sites
would thus provide input for calibration of existing models of budburst for mature forest
trees, as opposed to young trees in phenology gardens (see Appendix 1, section 3.0). This
objective should be taken into account when assessing the protocol. In addition, the data were
compared with girth band data to identify whether automatic girth bands are a suitable
indicator of budburst, to enable remote sensing of this critical phase of development.
The protocol described below was based upon the standard intensive phenology protocol for
Level II plots using a slightly modified scoring system. Ten selected numbered trees within
the Level II plot and the area immediately adjacent to the meteorological mast were
monitored on a daily basis from the ground using 8 x 30 binoculars. The frequency of
recording was reduced to twice weekly during the slow maturation of leaves in stage 4 and 5
in late May. The selected trees included the sample trees for biennial foliar analysis, the trees
used for intensive foliar analysis in 1998, and the four trees at each site that were
instrumented with girth bands. Monitoring extended from mid-April to the end of May commencing before bud burst until every tree attained full canopy development. Each tree
was assessed from the same point on each site visit, from which a large proportion of the
crown was visible. Phenological assessment was recorded on form 11b (Appendix II)
according to simplified instructions and photographs from the International Phenology
Garden Project (Appendix III). Assessment aspect was recorded on form 11c (Appendix II).
At Alice Holt, the feasibility of assessing intermediate stages between dormancy and the
appearance of green tissue was investigated and recorded separately on a modified version of
form 11b (11b*). Any damage was recorded on the standard Phenological assessment form
(11a; Appendix II).
Original definition of developmental stages:
• dormant
• green tissue visible
• definite leaf forms visible
• leaves fully expanded
Modified developmental stages:
• dormant
• buds expanded and probably broken – this stage can be assessed on the basis of the
shape or silhouette of the bud tip (a ‘ragged’ tip, rather than a pointed tip indicating that
this stage has been reached)
• green tissue visible
• definite leaf forms visible as defined by the appearance of petioles
• leaves fully expanded, as defined by canopy colour
A number of procedural difficulties were encountered, which are listed below, including
suggested remedies, which have been included in the modified definitions of developmental
stages, above:
1. An inability to distinguish between swollen buds and the point at which green tissue
became visible. This was particularly the case when assessing from beneath the crown (as
is the case out of necessity in dense stands), and also under overcast conditions; direct
14
sunlight increased the perception of colour. One solution would be to incorporate an extra
stage defined as swollen buds with ‘ragged’ tips, indicating that the bud scales were
parting.
2. The period represented by stage 2, (from bud burst to full leaf) was extended in many
cases, and the point at which stage 4 was attained was difficult to distinguish. The
definition of stage 4 having visible petioles provided a clearer separation between stages 3
and 4.
3. Difficulty in determining the day of complete leaf expansion (stage 5) for the most
advanced trees without any reference material in the rest of the canopy- trees developing
later were easier to score by comparison. Leaves are less vulnerable to frost damage at this
stage, and thus within the context of the objective of this procedure, the timing of the
transition between stages 4 and 5 is not critical. As an alternative, the transition between
stages could be defined on the basis of colour and attitude - stage 4 are light-mid green
and angled slightly upward; stage 5 are dark green and flattened to the horizontal.
4. The transition between stages 4 and 5 in the progress of flushing is difficult to distinguish
since it is unlikely that one stage is completed before the next begins higher in the crown.
Furthermore, some twigs/branches are dead, which only becomes apparent during the
latter stages of flushing. Progress Stage 5 should only be used to indicate completion of
developmental stage 5.
2.3.3 Foliar chemistry
At each site, five trees were selected on the basis of proximity to the meteorological mast to
gain access to the crown. During the summer of 1998, the foliage was sampled on a monthly
basis using squirrel poles, commencing after flushing was complete in late May. For each
tree, samples were collected from the top of the crown (exposed to full sunlight), from mid
crown, and from the base of the crown. Each sample consisted of a minimum of twenty
undamaged leaves which were labelled, and transported in sealed polythene bags.
Laboratory analysis consisted of the determination of specific leaf area (SLA: area per unit
weight) on an individual leaf basis using a leaf area analyser (CI-201: CID, Idaho, USA) and
four figure electronic balance). SLA was expressed as both fresh weight and dry weight (after
drying at 70oC for 48 hours) ratios. Kjeldahl nitrogen content was then assessed using
standard documented procedures, with N measured as NH4+ by colorimetric determination
using continuous flow analysis (Chemlab Instruments Ltd., Hornchurch, UK).
Identical sampling was also carried out on one occasion during August 1999 in order to make
an inter-annual comparison. Data were also compared with the routine biennial foliar analysis
carried out during August 1998, using conventional climbing and sampling techniques, to
provide an indication of errors that may be introduced into process modelling through the
existing foliar sampling techniques.
2.3.4 Litter collection
Ten litter traps were installed at each site during the summer of 1998. The sampling strategy
mirrored that of throughfall collection, with collectors offset by 2 m from the throughfall
collectors. The collectors consisted of green fibreglass funnels with sealed inner walls (60 cm
deep; 0.63 m2 area). A nylon mesh bag was attached to the spout of the funnel (15 cm
diameter) using a rubber ‘O’-ring. The litter bags were replaced by fresh bags and returned to
the co-ordinating laboratory complete for sorting and analysis as described below. Collections
were made every four weeks, except during heavy leaf-fall, when this period was reduced to
two weeks. Care was taken not to compress the leaves during transport to prevent damage to
15
material required for SLA determination. Compression of the samples also prevented the
accurate determination and sorting of other components of the litter sample.
Laboratory procedure
Live insects were removed as they should not be considered as litter, and to prevent
contamination of leaf samples (particularly for nitrogen analysis). Dead insects were
considered as litter, but were removed and weighed separately to plant material.
1. The litter bag was emptied onto a large plastic tray, and any live fauna removed, recorded
and released. Wet samples (as a result of rainfall) were allowed to air dry at room
temperature in a draught-free environment.
2. All leaf material was removed and oven-dried flat at 70 oC for 48 hours.
3. The remaining litter was sorted as twig/branch, bark, moss and lichen, seeds and flower
material (including bud scales) and dried as above.
4. All components were weighed on an individual collector basis.
5. At the end of leaf-fall, leaf area index of the stand was determined through an analysis of
bulked leaf samples from the entire leaf fall period. Two sampling strategies were
employed in the two years (1998 and 1999) to provide information on the variability
between collectors, and the temporal distribution of SLA of the litter. In 1998, samples
from all ten collectors were bulked by collection period, and 100 leaves from each bulked
sample period were chosen at random from a large mixing tray. In 1999, samples from all
sample periods were bulked on an individual collector basis, and 100 leaves from each
bulked collector sample assessed for SLA.
6. Specific leaf area of dried leaf samples was determined by weighing each sample of 100
leaves, re-wetting for 24 hours in shallow trays of water, and measuring the total area with
a leaf area meter fitted with a motor-driven sampling belt (Delta-T Devices, Cambridge
UK).
2.3.5 Derivation of light extinction coefficient (Kext) and leaf area index (LAI)
During the course of 1997, 1998 and 1999, canopy light transmission was measured above the
throughfall collectors (1997, 1998) and litter traps using two CEP40 ‘ceptometers’ (Delta-T
Devices, Cambridge, UK). One was operated from a forest clearing, and readings coincided
with those taken in the forest plot at predefined time intervals. Measurements were only made
on overcast days (zero beam fraction) such that the influence of leaf orientation was absent.
The effective light extinction coefficient (Kext) was calculated according to the relationship
described in the Beer-Lambert Law, with corrections made for light interception by woody
biomass:
Qtrans = Qinc exp((-Kext L)+b)
eqn. 1
where Qtrans and Qinc and transmitted and incident photosynthetically active radiation (PAR)
respectively, L is leaf area index (LAI) and b is the light extinction associated with woody
biomass.
2.3.6 Soil moisture
Soil moisture at 30 cm was measured using one theta probe (ML1: Delta-T Devices,
Cambridge, UK) at each site. Soil specific calibrations were obtained for both sites, using
soils from the relevant horizons (Fig. 5a&b). A calibration curve was also obtained for the
organic top soil at Grizedale. Triplicate soil cores (7.5 cm diameter, 7 cm deep) were obtained
for each soil and gradually oven dried at 80oC over the course of four days, and probe output
measured at regular intervals.
16
(a)
0.6
soil moisture content (v/v)
0.5
0.4
0.3
0.2
0.1
0
0
0.2
0.4
0.6
0.8
-0.1
theta probe output (V)
(b)
soil moisture content (v/v)
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.2
0.4
0.6
0.8
1
-0.1
theta probe output (V)
Fig. 5 (a) Theta probe calibration curve for the 30 cm soil horizon (C horizon; solid line) at Alice Holt. (b) Theta
probe calibration for the organic topsoil (open symbols, bold line) and mineral subsoil (closed symbols, fine line)
at Grizedale. In both cases, the default calibrations for mineral soils (dotted line) and organic soils (dashed line)
are shown.
17
3 Results
3.1 Meteorology
Manual climatological measurements continued throughout the duration of the project, with
100% data capture at Alice Holt, and only one weeks missing data at Grizedale. Table 1 gives
a summary of the meteorology over the period of the project at both sites.
(a)
month
Jan 1998
Feb 1998
Mar 1998
Apr 1998
May 1998
Jun 1998
Jul 1998
Aug 1998
Sep 1998
Oct 1998
Nov 1998
Dec1998
Jan 1999
Feb 1999
Mar 1999
Apr 1999
May 1999
Jun 1999
Jul 1999
Aug 1999
Sep 1999
Oct 1999
Nov 1999
Dec1999
Jan 2000
Feb 2000
Mar 2000
Apr 2000
May 2000
Tmax (oC)
8.3
10.8
11.5
12.0
19.3
18.6
19.9
22.7
19.4
14.4
9.2
8.9
8.6
8.1
11.5
14.0
18.1
19.5
23.7
21.7
20.4
15.0
10.6
8.0
7.9
9.8
11.4
11.7
17.5
Tmin (oC)
1.0
1.4
3.4
3.3
7.1
9.2
10.0
8.9
9.5
6.3
2.0
2.6
2.4
1.7
3.0
4.4
8.0
8.3
11.4
10.9
10.3
5.2
3.8
1.4
0.8
2.2
3.0
3.3
7.8
Ts (oC)
5.8
5.8
8.0
9.7
14.5
16.2
17.6
18.1
16.4
13.1
8.2
6.5
6.2
5.5
7.5
10.3
13.9
16.7
18.8
18.1
16.8
12.5
9.3
6.5
4.9
6.0
7.5
8.6
14.4
rainfall (mm)
110
8
70
107
24
97
37
30
91
153
69
89
139
36
41
50
74
70
7
84
113
77
36
131
25
79
29
158
94
run of wind (km)
4514
2798
3715
3384
2408
2970
2857
2186
3463
3462
2466
3967
4689
3759
3231
3745
3039
2265
2362
1808
2318
2582
3323
4491
3487
4222
3410
3613
3101
sunshine duration (h)
61
112
75
108
222
136
151
252
137
93
73
35
48
83
124
154
151
197
233
171
144
157
73
67
69
95
110
119
178
Tmax (oC)
7.0
9.4
10.3
10.9
17.5
15.9
16.8
17.4
17.4
12.3
8.3
8.0
7.6
7.4
9.1
12.2
15.8
16.6
20.1
19.0
18.0
13.2
10.5
6.8
7.1
7.7
10.0
10.2
16.4
Tmin (oC)
1.3
4.0
3.2
2.5
6.8
9.1
10.5
10.5
8.6
5.2
1.9
1.3
1.6
2.0
3.2
3.9
7.5
8.1
11.9
10.3
8.9
5.3
4.0
0.5
0.2
2.9
2.8
2.9
5.4
Ts (oC)
5.7
6.4
7.0
8.2
11.8
13.5
15.3
15.9
14.9
11.2
7.2
5.5
4.6
4.5
6.1
8.8
12.3
14.3
16.7
16.5
15.2
11.7
8.8
5.4
4.2
5.2
6.5
7.3
12.2
rainfall (mm)
171
155
229
115
91
230
135
224
107
394
206
225
315
102
161
127
152
122
90
90
128
241
150
385
243
297
183
95
91
run of wind (km)
2777
5142
3295
3056
2595
3112
2382
3091
2285
3976
2536
3503
4559
4183
3561
3525
3390
2686
2585
2107
2282
2607
3633
5016
4007
5916
3821
2512
2685
sunshine duration (h)
(b)
month
Jan 1998
Feb 1998
Mar 1998
Apr 1998
May 1998
Jun 1998
Jul 1998
Aug 1998
Sep 1998
Oct 1998
Nov 1998
Dec1998
Jan 1999
Feb 1999
Mar 1999
Apr 1999
May 1999
Jun 1999
Jul 1999
Aug 1999
Sep 1999
Oct 1999
Nov 1999
Dec 1999
Jan 2000
Feb 2000
Mar 2000
Apr 2000
May 2000
Table 1. Comparison of climatological data at (a) Alice Holt and (b) Grizedale for the duration of the project.
18
The differences between sites are represented graphically in Fig. 6a-c.
(a)
air temperature ( oC)
25
20
15
10
5
0
Jan-98
Jul-98
Feb-99
Aug-99
Mar-00
(b)
450
monthly rainfall (mm)
400
350
300
250
200
150
100
50
0
Jan-98
Jul-98
Jan-99
Jul-99
Jan-00
(c)
monthly run of wind (km)
7000
6000
5000
4000
3000
2000
1000
0
Jan-98
Jul-98
Feb-99
Aug-99
Mar-00
Fig. 6. Comparison of (a) average monthly maximum and minimum temperatures, (b) monthly rainfall and (c)
monthly run of wind at Alice Holt (bold lines, solid bars) and Grizedale (fine lines, clear bars).
Minimum temperatures and mean windspeed are similar at the two sites. However, summer
maximum temperatures are higher at Alice Holt compared to Grizedale, and Grizedale has
over three times the annual rainfall of Alice Holt. These climatological data have been used to
derive hourly data for input to the process model (see methods and modelling sections). A
limited comparison of these climatological data and data from the canopy weather station is
possible through the minimum and maximum temperature readings, and also soil temperature
and vapour pressure at 9 am. These data are presented in Fig. 7a-h.
19
(a)
(e)
15
'manual' T min ( C)
( C)
15
10
min
o
o
'manual' T
5
0
-5
0
10
10
5
0
-5
0
5
10
15
-5
-10
canopy Tmin (oC)
canopy T min ( oC)
(b)
(f)
30
25
'manual' T max (oC)
'manual' T max ( C)
30
25
o
20
15
10
5
20
15
10
5
0
0
10
0
20
0
o
canopy Tmax ( C)
25
25
20
20
15
10
5
15
10
5
0
0
0
5
10
canopy Ts ( oC)
15
20
0
(d)
5
10
canopy Ts ( oC)
15
20
(h)
25
'manual' vapour pressure
(mbar)
25
'manual' vapour pressure
(mbar)
30
(g)
'manual' T s ( oC)
'manual' T s ( oC)
(c)
10
20
canopy Tmax ( oC)
20
15
10
5
0
20
15
10
5
0
0
5
10
15
20
canopy vapour pressure (m bar)
25
0
10
20
canopy vapour pressure (mbar)
Fig. 7. Comparison of meteorological variables from ‘manual’ climatological and canopy automatic weather
stations at (a-d) Alice Holt and (e-h) Grizedale. Lines are fitted by linear regression (see Table 2).
20
30
variable
Alice Holt
slope
intercept
‘manual’ climatological vs canopy AWS comparison
Tmin
0.98
-0.93
Tmax
1.05
+0.12
Ts
1.37
-1.25
vp
0.95
+0.34
r2
Grizedale
slope
intercept
r2
0.97
0.99
0.97
0.95
1.08
1.02
1.35
0.96
-1.64
+0.66
-1.1
+0.26
0.86
0.97
0.96
0.93
climatological vs canopy AWS comparison
Tmin
Tmax
Ts
vp
1.002
1.03
1.30
0.97
+0.23
0.45
-1.02
0.19
0.98
0.99
0.97
0.95
Table 2. Summary of relationships between manual climatological and canopy automatic weather station data for
Alice Holt and Grizedale, and a comparison of two automatic weather stations, one within the plot, and the other
adjacent to the climatological station.
At both sites, the comparison between manual climatological and canopy automatic weather
station variables was favourable. Although vapour pressure showed some scatter, the slope
and intercept (Table 2) suggest that vapour pressure differences are insignificant over the
spatial scale represented at the two sites. However, minimum temperature was consistently
lower as measured by the climatological station. This may be a result of instrumentation and
screen differences, or representative of different climatic conditions. The comparison of
automatic weather station data at the forest plot and at the site of the climatological station
(Alice Holt; Fig. 8) suggest that the difference is largely a result of differences in
instrumentation, since the intercept was closer to zero than in the canopy – climatological
station comparison. However, the larger scale of the differences at Grizedale indicates that the
microclimate of the two stations may differ, which is not surprising since one is on a valley
floor, and the other above the canopy on a hillside. The larger scatter at Grizedale is indicative
of these differences, and suggests that minimum temperature is an important variable to
measure at the forest plot, particular because of its importance in predicting frost damage.
30
10
o
25
AWS Tmax (oC)
AWS Tmin ( C)
15
5
0
0
20
15
10
10
5
-5
0
0
canopy Tmin (oC)
10
20
o
canopy Tmax ( C)
25
AWS vapour pressure (mbar)
25
AWS T s ( oC)
20
15
10
5
0
0
5
10
15
20
20
15
10
5
0
0
canopy Ts ( oC)
5
10
15
20
25
canopy vapour pressure (m bar)
Fig. 8. Comparison of automatic weather station (AWS) data for the forest plot and climatological station at
Alice Holt. Data represent daily values (minimum or at 9 am) for the duration of the project.
21
Soil temperature showed the largest deviation from unity, both between canopy AWS and
climatological station data at Grizedale and Alice Holt (Table 2), and also in the relationship
between the two AWS datasets at Alice Holt (Figs. 7&8). This suggests that this is not a result
of instrumentation, which is supported by the annual distribution of these differences, with Ts
at the open climatological station being 3-4oC warmer on average during the summer months,
but with little difference during the winter (Fig. 9a&b). The difference between the open and
canopy sites is a result of differences in radiation balance between the two sites, and its
magnitude may assume great importance when modelling soil processes including respiration.
(a)
(b)
25
20
soil temperature (oC)
soil temperature (oC)
18
20
15
10
5
16
14
12
10
8
6
4
2
0
0
Jul-98 Dec-98 May-99 Oct-99 Mar-00
Jul-98 Dec-98 May-99 Oct-99 Mar-00
Fig. 9. Seasonal course of soil temperature at the climatological station (open symbols) and canopy AWS (closed
symbols at (a) Alice Holt and (b) Grizedale.
22
3.2 Canopy development
The objective of this section of the project was to devise and demonstrate a methodology that
would provide information on canopy development at key phases, providing additional input
for modelling activities in a number of areas:
• The timing of budburst, including susceptibility to frost damage; this may be of particular
relevance given predictions of a more variable climate, and for a generally warmer climate
leading to advancement of budburst
• The rate of development of leaf area (for broadleaf species) which is an essential input for
process driven models of forest growth; canopy development also impacts on the
calculation of water budgets, both through its effect on transpiration water losses, and also
through the magnitude of interception losses
• Maximum leaf area index; again an essential input for process models, and will also
provide data for comparison with crown condition assessments
3.5
3
LAI
2.5
2
1.5
1
0.5
0
0
50
100
150
200
day of year
250
300
350
Fig. 10. Preliminary assessment of LAI at Alice Holt using Ceptometers in 1997. In the absence of litter fall
data, LAI was calculated assuming a default value of 0.82 for Kext
Preliminary work in 1997 demonstrated that under overcast conditions, canopy light
transmission could indicate the relative development of canopy development (Fig. 10).
However, no data were available to derive a value for the effective light extinction coefficient,
and thus calibrate the Beer-Lambert law. In the absence of these data, a default value of 0.82
was assumed for Kext, representing leaf level light extinction rather then effective canopy light
extinction. The low estimates for LAImax (3.1 vs 6.4 in 1998) highlights the importance of
deriving an accurate value for effective Kext. Measurements of canopy light transmission were
again made in 1998 within the level II plot, with sample points coinciding with throughfall
collectors. Litter traps were installed in summer 1998 and litter fall data for 1998 and 1999
are shown in Table 3.
Site
Alice Holt
Year
1998
1999
Oak
Weight
(g m-2)
296
324
SLA
(cm2 g-1)
189
172
Area
(m2 m-2)
5.59
5.58
Grizedale
Ash/others
Weight
(g m-2)
40
35
LAI
SLA
(cm2 g-1)
-
Area
(m2 m-2)
0.64
0.83
1998
210
137
2.89
3.1
1999
237
133
3.15
0.2
Table 3. Litter fall data used to estimate LAI at Alice Holt and Grizedale in 1998 and 1999.
6.23
6.41
2.89
3.15
Leaf area index data shown in Table 3 were then integrated with mean canopy transmission
data for the two level II plots, measured using Ceptometers (Delta-T Devices, Cambridge,
UK) shown in Table 4 to derive effective light extinction coefficients for the two canopies.
23
site
date
Alice Holt
τ
mean PAR
µmol m-2 s-1
374
224
282
211
05/05/98
26/08/98
02/10/98
03/08/99
LAI
0.497
0.086
0.086
0.086
Kext
0
6.23
6.23
6.41
0.39
0.39
0.38
Grizedale
17/08/98
102
0.220
2.89
0.52
05/10/98
84
0.280
2.89
0.44
25/08/99
52
0.320
3.15
0.36
Table 4. Derivation of Kext from measured canopy light transmission (τ) above throughfall collectors at Alice
Holt and Grizedale in 1998 and 1999.
It should be noted that the values for extinction coefficients given in Table 4 are for effective
extinction coefficients during the period of complete canopy cover. An error is therefore
introduced when canopy cover is not complete, as shown in Figure 11, where winter LAI
assumes a value of approximately 0.6. In absolute terms, this error amounts to approximately
10% of full canopy cover at Alice Holt, and 20% at Grizedale. Although the errors are
significant, when the extinction coefficients are applied to continuous light transmission data
using solarimeters mounted above and below the canopy using the meteorological mast, the
protocol does provide an accurate method for monitoring canopy development as shown in
Figs. 12 and 13. This monitoring of canopy development is particularly successful during the
period of rapid canopy development in spring and early summer. It also demonstrates the
continuing canopy development throughout the summer at Alice Holt, a process that was not
observed at Grizedale. A comparison of the three years during which measurements were
made is given in Fig. 11a&b.
(a)
6
1998
5
1999
LAI
4
2000
3
2
1
0
0
50
100
150
200
250
300
350
400
day of year
(b)
LAI
7
1998
6
1999
5
2000
4
3
2
1
0
0
50
100
150
200
250
300
350
400
day of year
Fig. 11. Comparison of leaf area development for Alice Holt (a) and Grizedale (b) in 1998, 1999 and 2000 as
derived from measurements of canopy light transmission.
24
At both sites, total leaf area development (to date) in 2000 has been considerably larger than
that in 1998 or 1999 (Fig. 11), possibly as a result of climatic conditions, particularly the wet
spring experienced in 2000. At Alice Holt, it may also represent the closing of the canopy
following the severe thinning operation that took place in 1996 at the site of the
meteorological tower. This hypothesis is supported by the (small) increase in LAI that was
observed in 1999, compared with 1998. At Grizedale, canopy development in 1998 and 1999
was almost identical.
(a)
5
4
LAI
3
2
1
0
0
50
100
150
200
250
300
day of year
(b)
7
6
LAI
5
4
3
2
1
0
50
100
150
200
250
300
200
250
300
day of year
(c)
6
5
LAI
4
3
2
1
0
50
100
150
day of year
Fig. 12. Development of leaf area index derived from measurements of canopy light transmission at Alice Holt.
a) 1998; b) 1999; c) 2000.
25
(a)
7
6
LAI
5
4
3
2
1
0
0
50
100
150
200
250
300
350
400
250
300
350
400
day of year
(b)
7
6
LAI
5
4
3
2
1
0
0
50
100
150
200
day of year
(c)
6
5
LAI
4
3
2
1
0
0
50
100
150
200
250
300
350
400
day of year
Fig. 13. Development of leaf area index derived from measurements of canopy light transmission at Grizedale. a)
1998; b) 1999; c) 2000.
Phenological development was also monitored on a daily basis (up to three days during slow
moving phases) in spring 2000. Details of the protocol are given in section 2.3.2 and
Appendices II and III. Mean canopy development score is compared with canopy
development as derived from canopy light transmission (Fig. 14c), as described above, and
the phenological record is also compared alongside the automatic girth monitoring data (see
section 3.3).
26
(a)
Relative development
1.2
light transmission
1
phenology
0.8
0.6
prediction
0.4
0.2
0
90
100
110
120
130
140
150
160
140
150
160
day of year
(b)
Relative development
1.2
light transmission
1
prediction
phenology
0.8
0.6
0.4
0.2
0
90
100
110
120
130
day of year
Fig. 14. Comparison of spring canopy development measured by canopy light transmission and phenological
assessment for (a) Alice Holt and (b) Grizedale during spring 2000. The dates of budburst predicted by the
synthesis model of Hanninen (1990) are also shown. Relative development is assessed as the product of the
numerical stage of development and the number of trees attaining that stage, relative to the maximum value for
this index (ie 1.0 represents complete canopy development).
The two methods of assessing canopy development vary greatly in both their timing and rate
of apparent development. Both datasets provide invaluable data inputs for process modelling.
Measurements of light transmission do not indicate the date of budburst, which is provided by
standard phenological observations. The timing of budburst is essential for predicting any
adverse effects of late spring frost. However, the phenological assessment used in this study
overestimates the rate of canopy development, whether the stage of development, or progress
of flushing is used as the measure of development. Phenological monitoring will overestimate
canopy photosynthesis, transpiration and rainfall interception if used as a model input. The
synthesis model of Hanninen (1990) has also been used to predict the date of budburst from
meteorological data. The model was parameterised using daily climatological data (19691981), and thus those datasets have been used for the model simulations rather than the stand
specific AWS data. The prediction of budburst agrees well with observed budburst at Alice
Holt, which is encouraging, since the model was parameterised for this site (see Appendix I;
model description and validation). At Grizedale, the predicted and observed date of budburst
differed by approximately seven days. A number of hypotheses could explain these
differences:
27
•
•
•
•
poor model performance; the model parameters were the same as those for Alice Holt,
with only latitude and meteorological inputs changing
discrepancy between weather conditions at the forest plot and the meteorological station;
although the two weather station were only 2 km apart, observed minimum temperatures
differed by up to 10oC as a result of topographic effects. This difference in minimum
temperature between the two weather stations was not evident at Alice Holt
an experimental artefact; the crown at the Grizedale plot was lower than that at Alice Holt,
and thus the early stages of flushing were easier to distinguish
species differences which were not reflected in the model parameterisation; Alice Holt
consists of a mixture of Q. petraea and Q. robur (and some Q. cerris), whilst climate
limits Grizedale to Q. petraea.
The linear regressions of within plot AWS and climatological station daily minimum
temperature suggest that the discrepancy between modelled and observed budburst at
Grizedale could indeed be a result of temperature inputs to the model, since the x-axis offset
at Grizedale was -1.6oC compared with only -0.9oC at Alice Holt (Table 2).
28
3.3 Girth increment as an indicator of budburst
Since the precise timing of budburst can be crucial to the impact of late spring frost, an
attempt was made to monitor this process remotely, using automatic girth bands. The general
performance of the girth bands is shown in Fig. 15, and demonstrates that annual girth
increment can successfully be monitored using the D6 strain gauges. The periods of rapidly
fluctuating sensor output at both sites are a result of poor data-logger performance prior to
failure.
(a)
girth increment (mm)
60
50
40
30
20
10
Jan-98
Jul-98
Feb-99
Aug-99
Mar-00
Oct-00
Aug-99
Mar-00
Oct-00
date
(b)
girth increment (mm)
50
45
40
35
30
25
20
Jan-98
Jul-98
Feb-99
date
Fig. 15. Course of automatic girth band output for the duration of the project for (a) Alice Holt and (b) Grizedale.
Individual tree girth increment data are shown for spring 2000 in Fig.16 together with
observed flushing dates for both Alice Holt and Grizedale. This study suggests that girth
bands do not provide a reliable indicator of flushing in oak as has been suggested by some
authors.
There is a suggestion that at Grizedale, stem shrinkage was observed at the time of leaf
expansion (from day 120). However, this response was confounded by relationships with soil
moisture (or possibly bark hydration following rain events) and temperature as shown in Fig.
17. Under dry climatic conditions, this technique may indicate budburst, but under the
climatic conditions of the UK, automatic girthbands cannot be used to reliably indicate
budburst. Future studies may find more success if the stem above and below the band is
protected from rain, thereby discounting the possibility of bark hydration interfering with the
signal of budburst.
29
50
60
(a)
(b)
55
45
girth increment (mm)
girth increment (mm)
50
45
40
40
35
35
30
30
25
4-Apr
14-Apr
24-Apr 4-May
date
14-May
25
4-Apr
24-May
14-Apr
24-Apr
4-May
date
14-May
24-May
Fig. 16 Comparison of automatic girth band data with observed date of budburst in 2000 for a) Alice Holt, and b)
Grizedale. Data are shown for individual trees. Arrows indicate the observed date of flushing (corresponding to
stage 5 in form 11b).
(a)
(c)
48
53
girth increment (mm)
girth increment (mm) or soil moisture
content (%v/v)
55
51
49
47
46
47
45
45
60
70
80
90
48
100
(d)
50
54
32
31
45
girth increment (mm)
girth increment (mm) or soil moisture
content (%v/v)
52
soil m oisture (% v/v)
day of year
(b)
50
40
35
30
29
28
30
27
25
60
70
80
90
35
100
40
45
50
soil m oisture (% v/v)
day of year
Fig. 17 (a&b) Girth increment (bold lines) and soil moisture content (fine lines) for March 2000 at Alice Holt
and Grizedale, respectively. (c) and (d) relationship between girth increment and soil moisture content from a&b.
In each case, girth increment data are shown for a single representative tree to aid clarity.
30
The confounding relationship between soil moisture content and girth increment is shown for
March 2000 at both sites (Fig. 17c&d). These data are representative of the dormant period,
and thus any girth changes are the result of physical rather than physiological changes. The
weather patterns experienced at the two sites differed markedly over this period, with the
rainfall for March 2000 amounting to 120 mm at Grizedale, and only 25 mm at Alice Holt.
These differences are reflected in the patterns of soil moisture content at 30 cm, but in both
cases, relationships between soil moisture content and girth increment are apparent. The
strange form of the relationship at Grizedale may indicate a lag in the response between soil
rehydration and change in girth, a hypothesis that is supported by the tighter data set at Alice
Holt, where continuous soil drying was observed.
31
3.4 Foliar Chemistry
The relationships between foliar nitrogen concentration ([N]) and the photosynthetic
parameters Jmax and Vmax are well described in the scientific literature (Medlyn et al., 2000;
Kull and Jarvis, 1995). Specific relationships for oak were derived at Alice Holt in 1996 (Fig.
18). This piece of work demonstrated large variation in Jmax and Vmax both through the course
of the growing season, and also with position in the canopy (ECOCRAFT, 1999; Table 5).
Since Vmax and Jmax can be estimated through routine analysis of foliar N, it is essential that a
suitable sampling protocol is adopted, which can reflect developmental changes in foliar
chemistry that have a bearing upon simulated annual productivity.
J max or V max (µ mol m -2 s -1)
250
200
150
100
50
0
0
1
2
3
4
5
-2
foliar [N] a (g m )
Fig. 18 Relationship between the photosynthetic parameters Jmax (open symbols: Jmax=49.9[N]a+17.3; r2=0.83)
and Vmax (closed symbols: Vmax=16.9[N]a+8.2; r2=0.83) and foliar nitrogen content expressed as a function of
area.
leaf respiration rate ( µ mol m -2 s -1)
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0.5
1
1.5
2
2.5
foliar [N] a (g m -2)
3
3.5
Fig. 19 Relationship between leaf respiration at 25oC and foliar nitrogen content, both expressed on an area basis
(Ra=0.607[N]a-0.049; r2=0.83).
Foliar samples were collected from five trees within reach of the meteorological mast using
squirrel poles during the 1998 growing season (Fig. 20). At Grizedale, the first sampling point
(19 May) was prior to full canopy development, and as a consequence, [N] expressed as a
function of weight was higher than for other sampling periods. At Alice Holt, the first
sampling point was two weeks later than at Grizedale (2 June), and canopy development was
nearly complete, with values of [N] typical of the rest of the growing season observed. At
32
both sites, [N] at the final sampling point (7 November) was considerably lower, reflecting
reallocation of [N] prior to autumn leaf fall. Throughout the rest of the growing season, there
was little variation in [N] and no relationship with canopy position was apparent. Samples
were also collected at a single time-point during the 1999 growing season, yielding almost
identical results to 1998. Results are also compared with those from routine biennial foliar
sampling in Table 5. These data suggest that at Alice Holt, canopy access using a scaffold
tower and squirrel poles provided foliage more representative of the canopy surface than at
Grizedale where access was obtained through tree climbing and the use of squirrel poles.
year
Alice Holt
Grizedale
middle
bottom
top
middle
bottom
1998
2.49
2.44
2.08
2.2
2.16
1999
2.46
2.59
2.01
2.19
2.15
1999
2.31
2.15
Table 5 Comparison of [N] between 1998 and 1999, and between targeted and routine sampling protocols in
1999.
top
2.39
2.32
Absolute values of [N] were lower at Grizedale throughout the period of the study (Fig. 20),
which may be a result of a smaller solar radiation input than at Alice Holt (Kull and
Niinemets, 1998; Kull and Kruijt, 1999), or may indicate a different nutrient status, either
nitrogen limitation, or some other limiting factor; phosphate deficiency is a possibility, since
total phosphorus at Grizedale was only half that at Alice Holt (see Figure 21), and was close
to the definition of P-deficient (deficient<0.14, optimum>0.16; Taylor, 1991).
3.5
foliar [N] (%)
3
2.5
2
1.5
1
100
150
200
250
day of year
300
350
Fig. 20 Variation in foliar nitrogen content with time of year at Alice Holt (solid symbols) and Grizedale (open
symbols). The upper canopy is represented by circles and the lower canopy by triangles in each case.
0.3
foliar [P] (%)
0.25
0.2
0.15
0.1
0.05
0
150
200
250
day of year
300
350
Fig. 21 Variation in foliar phosphorus content with time of year at Alice Holt (solid symbols) and Grizedale
(open symbols) during 1998 (circles) and 1999 (squares).
However, when foliar nitrogen content is expressed as a function of area by adjusting [N] for
differences in specific leaf area (SLA: cm2 g-1), considerable variation, both in terms of time
33
of year and canopy position, becomes apparent (Fig. 22). This variation is highly significant,
since most models of photosynthesis are variations of the model of von Caemmerer and
Farquhar (1981) which expresses the process on a leaf area basis, as does the parameter
derivation used here (see Figs. 18 and 19).
-2
foliar [N]a (g m )
3.5
3
2.5
2
1.5
1
0.5
100
150
200
250
day of year
300
350
Fig. 22 Variation in foliar nitrogen content expressed as a function of area for upper (circles) and lower
(triangles) canopy positions at Alice Holt (closed symbols) and Grizedale (open symbols) in 1998.
This considerable variation in [N]a is largely a function of variation in SLA, and if data points
are restricted to those representing a fully developed canopy and prior to autumn senescence,
a tight relationship is observed between [N]a and SLA (Fig. 23).
3.5
foliar [N]a (g m-2)
3
2.5
2
1.5
1
0.5
50
100
150
specific leaf area (cm 2 g-1)
200
Fig. 23 Relationship between foliar nitrogen content expressed as a function of area and specific leaf area for
Alice Holt (closed symbols, bold line: [N]a=1e-4(SLA)2-0.045(SLA)+5.72; r2=0.85) and Grizedale (open
symbols, fine line: [N]a=6e-5(SLA)2-0.027(SLA)+4.21; r2=0.85) for all samples collected June-October 1998 and
August-September 1999. Regressions are fitted by least sum of squares assuming quadratic relationships.
The different relationships observed in Fig. 23 may be as a result of species differences (Alice
Holt is a mixture of Q. petraea and Q. robur and their hybrids, whilst Grizedale is limited to
Q. petraea), or different nitrogen availability at the two sites, or to different solar radiation
environments, affecting the development of the photosynthetic apparatus and thus [N]a, since
RuBisCO comprises approximately 50% of leaf protein. The tightness of the site specific
relationships shown in Fig. 23 does indicate that after an initial ‘calibration’ period, specific
leaf area can be used as a surrogate for [N]a. However, further work is required to test this
approach, particularly across a range of sites subjected to varying N deposition.
34
3.5 Soil moisture
Soil moisture was measured both to act as a driver input to the process model, and also to act
as validation for the water balance budget of the model simulations. The following section
outlines the soil moisture measurements at the two sites, whilst the comparison with model
output is given in the modelling section for 1999 only.
soil moisture (% v/v)
(a)
100
90
80
70
60
50
40
30
20
10
0
Jan-98
Jul-98
0-15 cm
60-90 cm
Feb-99
Aug-99
15-30 cm
90-120 cm
Mar-00
Oct-00
30-60 cm
theta (30 cm)
.
(b)
70
soil moisture (% v/v)
60
50
40
30
20
10
0
Jan-98
60-90 cm
30-60 cm
Jul-98
0-30 cm
theta (30 cm)
Feb-99
Aug-99
Mar-00
Oct-00
Fig. 24 Seasonal course of soil moisture content for the duration of the project at a) Alice Holt and b) Grizedale.
In both cases, the 30 cm reading is from the continuous theta probe record, whilst all others are spot readings of
the 5 sector TDR probes. At Alice Holt, readings are the average of the two probes at each depth. Single probe
measurements are given for Grizedale.
Large seasonal fluctuations were observed in volumetric soil water content at Alice Holt (Fig.
24a), with field capacity approximately 50%, a value typical of the Denchworth series gley
(Hall et al., 1977). During the summer months in both 1998 and 1999, the moisture content of
the top 15 cm fell to about 10%, reflecting the higher organic content of the A horizon. The
35
water content of the B and C horizons did not fall below 25%, values again typical of soils
with a high clay content. The output from the TDR probes agreed well with the theta probe
output, both with depth (the closest fit was with the 30-60 cm sector) and with time. The very
high moisture content in the 90-120 cm sector is probably an experimental artefact due to
high compaction in the lower horizons of the sub soil. However, these values do indicate that
there was no significant moisture deficit at this depth, and that drainage from the subsoil was
minimal. In contrast, at Grizedale there was little evidence of seasonal patterns of soil water
content, with the soil moisture regime dominated by episodes of heavy rainfall and rapid
drainage (Fig. 24b). The rapid fluctuations at 60-90 cm is likely to be a function of the high
shale content at this depth and transitory drainage pulses with little water retention. Care was
taken when installing the theta probe to exclude all stones from the soil. The A horizon in the
vicinity of the weather station was unusually deep, and although at a depth of 30 cm, these
data are representative of the organic soil rather than the mineral sub-soil. The combination of
higher organic matter content and absence of stones provides an explanation for the generally
higher and less variable moisture content when compared with the TDR probe output at a
similar depth.
18
(a)
0-15 cm
16
15-30 cm
frequency
14
30-45 cm
12
45-60 cm
10
60-75 cm
8
6
4
2
0
15
20
25
30
35
40
45
50
moisture content (% v/v)
55
60
14
0-15 cm
frequency
(b)
12
15-30 cm
10
30-45 cm
45-60 cm
8
60-75 cm
6
4
2
0
5
10
15
20
25
moisture content (% v/v)
30
35
Fig. 25 Frequency distribution of soil moisture content by depth at (a) Alice Holt, and (b) Grizedale.
36
Soil moisture measurements were also made at one time point at positions corresponding to
the soil pits dug during the profile descriptions in 1996 as part of the level II soil survey.
Sampling points were offset by 1 m (to the north) to prevent soil disturbance from affecting
the results. The high stone content at Grizedale resulted in low measured moisture content,
partly as a result of poor contact between the probe and the soil, and partly because of the
resultant low moisture holding capacity. The frequency distribution by depth (Fig. 25)
indicates a normal distribution around 17% v/v for the upper 15 cm, with an abundance of
lower moisture contents at greater depths. Together with the large variation in soil depth that
was apparent at the time of the soil description, this survey indicates a highly variable soil at
Grizedale in terms of its water holding capacity. This variability is in contrast to Alice Holt,
where the deeper soil and absence of stones led to complete probe insertion at all locations.
The absence of stones also led more uniform moisture contents at a given depth, and a
consistent gradient of moisture with depth was observed. Some spatial variation was evident,
and is likely to represent variation in the depth of the organic horizons above the
predominantly clay sub-soil, and also, variation in drift over the Cretaceous clay. Drainage
ditches also ran through the Alice Holt plot and may have contributed to the variability. Soil
moisture measured in these spatial analyses did not differ significantly from either the
continuous record or the fortnightly TDR record, and indicates that the continuous record will
provide suitable data for model validation, if extended to cover both the organic and mineral
horizons.
The two soil types found at Alice Holt and Grizedale thus represent contrasting ends of the
soil spectrum, with both presenting difficulties for process modelling of soil water balance,
particularly the large variation in depth found at Grizedale. The uniformity of the soil at Alice
Holt does however, enable the soil moisture dataset to be used for model validation. The
highly variable nature of the soil at Grizedale precludes its use as a validation dataset for soil
water budgets (although spatial assessments of tree health and vitality and soil mapping are a
possibility), and the use of the measurements of soil moisture made in this study at Grizedale
will be restricted to the continuous theta probe record being used as an environmental driver.
37
3.6 Modelling
The modelling of long-term forest growth and yield requires more site and species specific
input parameters than are available from this project, and it is not the intention of this
modelling section to produce absolute predictions of forest growth. However, the results
obtained in the preceding sections do provide sufficient site specific parameters to enable the
modelling of annual carbon and water budgets. Furthermore, this approach allows a
comparison of the sensitivity of model outputs to variation in the various input parameters and
variables that have been derived or collected during the monitoring phase of the project. It
should also be borne in mind that many parameters are not site specific and can be obtained
from the scientific literature, although incompatible datasets can introduce errors into model
outputs.
The model simulations have been undertaken using the process model GROMIT (Growth
Model of Individual Trees) which is currently under redevelopment to become incorporated
within the Forestry Commission stable of growth, yield and environmental response models.
The code is written in Fortran 90 and is available to run on either Unix or PC platforms. The
major physiological processes are represented by conventional models [stomatal conductance:
Jarvis (1976); photosynthesis: von Caemmerer and Farquhar (1981); evapo-transpiration:
Monteith (1965); rainfall interception: Whitehead and Kelliher (1991); light interception:
Meng and Arp (1994)]. The model operates at an hourly time-step; if meteorological drivers
are only available as daily climatological data, a stand alone weather generator module
(running only on a Unix platforms) is also available to simulate hourly data. Budburst is also
predicted external to the main module, as temperature data for the preceding year are required
to fulfil the chilling requirement of the synthesis model of Hanninen (1990). Full
documentation of the models is given in Appendix I, including validation. Details of the
application of the model to climate change and rising atmospheric carbon dioxide
concentrations may be found in ECOCRAFT (1999), whilst the application of the model to
the effects of ozone pollution are given in Broadmeadow et al. (1999).
3.6.1 Approach
All results are expressed as net annual carbon (and water) flux for 1999, since 1999 is the
only complete year during which the project was running. Foliar chemistry data are only
available for 1998, and the parameters derived from them have been applied to the 1999
simulations. The sensitivity to each of the parameters derived in this project is tested, and the
output also compared with default values obtained either from the scientific literature, or from
other sources. The sensitivity of the model output to meteorological data is also tested,
including an assessment of the value of using plot specific meteorological data, and an
analysis of which variables have the greatest influence of model output, and thus provide a
ranking of desirability for within plot weather stations.
3.6.2 Baseline simulations
‘Baseline’ simulations of annual carbon fluxes have been prepared using all information from
field measurements made during the course of this project. The output from these simulations
is compared with the output simulated under various other scenarios in the sensitivity
analysis. The following section describes the derivation of input parameters which are not
given in earlier sections.
Parameters describing both the seasonal course of leaf area production (LAI) and the
development of photosynthetic capacity (Jmax and Vmax) have been derived using the approach
described in section 3.4 (Table 6); Jmax and Vmax are calculated using the 1998 foliar chemistry
data through applying the relationships derived in earlier work on oak (see Fig. 18). The
development of leaf area index has been quantified by assuming two linear phases (spring and
38
summer) fitted to the derived LAI data shown in Fig. 11, correcting for light interception of
woody material, and the difference between maximum LAI as represented in Fig. 11 and that
measured by litter collection (eqn. 2)
Lt =(t/tb-ta)*(L’b-L’a-L’o)*Lmax*(L’max-L’o)
eqn. 2
where Lt is leaf area index at time t, ta and tb are the beginning and end points of the
developmental phase encompassing t, L’a L’b and L’o and L’max are the uncorrected values for
L at ta, tb prior to budburst and after complete canopy development and Lmax is LAI measured
through litter collection.
Additional timepoints have been added by linear interpolation during the period of rapid
canopy development in spring/early summer, although photosynthetic capacity assumes the
same value as at the first point of foliar sampling. Rd (foliage dark respiration in the presence
of light) has also been derived from foliar nitrogen content using the relationship given in Fig.
19.
day
LAI
Alice Holt
121 0
126 0.75
131 1.45
141 2.91
153 4.65
188 5.42
216 5.77
245 6.16
260 6.41
280 0
SLA
(cm2 g-1)
Rd (µ
µmol m-2 s-1)
top middle bottom
Jmax (µ
µmol m-2 s-1)
top
middle bottom
Vmax (µ
µmol m-2 s-1)
top
middle bottom
143.5
143.5
143.5
143.5
132.1
123.3
128.2
130.0
0.60
1.23
1.23
1.23
1.61
1.70
1.53
1.33
61.0
98.0
122.5
122.5
153.9
161.1
147.2
130.4
22.0
35.5
43.8
43.8
54.5
56.9
52.2
46.5
0.93
0.93
0.93
1.06
1.10
1.06
1.07
0.78
0.78
0.87
1.00
0.97
0.96
85.5
98.0
98.0
108.4
111.6
108.1
109.7
85.5
85.5
92.8
103.4
101.3
100.5
31.3
35.5
35.5
39.0
40.1
38.9
39.5
31.3
31.3
33.8
37.4
36.6
36.4
Grizedale
116 0
121 0.75
211.9
0.52
53.0
14.0
126 0.83
211.9
0.69
77.8
28.7
132 1.38
211.9
0.85 0.69
91.0
77.8
33.2
28.7
139 1.92
211.9
1.04 0.85
106.4 91.0
38.4
33.2
167 2.88
117.7
1.11 1.15
0.85
112.9 115.9
90.8
40.6
41.6
33.1
195 2.94
119.6
1.39 1.25
0.84
136.0 123.9
90.7
48.4
44.3
33.1
223 3.01
127.2
1.24 1.34
0.77
123.6 131.1
85.0
44.2
46.7
31.1
251 3.08
116.3
1.32 1.06
0.72
129.7 108.5
80.2
46.3
39.1
29.5
280 0
Table 6 Parameters describing the seasonal development of LAI and photosynthetic capacity used as input for
model simulations. LAI is derived from canopy light transmission (Figs. 12&13), corrected for stem and branch
interception.
3.6.3 Respiratory components
The validation of model output given in Appendix I is with respect to ecosystem CO2 and
water vapour fluxes. Foliage respiration is modelled as a function of foliage mass calculated
from LAI and specific leaf area (SLA) at individual timepoints (see Table 6). Respiration rate
per unit leaf mass is in turn modelled as an exponential function of air temperature, assuming
a Q10 of 2.0. Foliage respiration rate at a reference temperature (25oC) was measured at five
positions in the canopy in 1997. No relationship with canopy position was observed, and thus
a single mean value of 14.3 nmol CO2 g-1 s-1 has been assumed for all canopy positions and
throughout the year in the model simulations. Root, stem and branch respiration is not
39
separated into the individual components, but is modelled as a single system respiration, again
as an exponential function of temperature, based upon the night-time (0000-0600 and 18000000 h) eddy correlation CO2 flux measurements in December, January, February and March
1998-99 restricted to windspeeds above 3 m s-1 thereby reducing the impact of canopy storage
and subsequent turbulent release of CO2 (Fig. 26).
-2
-1
CO2 flux (µ mol m s )
6
5
4
3
2
1
0
0
5
10
o
air temperature ( C)
15
Fig. 26 Relationship between night-time CO2 flux measured during winter by eddy correlation and air
temperature used in the derivation of the wood respiration – temperature function (y=0.91exp(0.087Ta); r2=0.39).
Although the eddy correlation flux site is 200 m distant to the intensive monitoring site, it is
assumed that the same system respiration – temperature relationship holds at both. No flux
data are available for deriving a system respiration function at Grizedale, and thus the
relationship observed at Alice Holt was corrected for the difference in biomass at the two
sites, using timber volume from the mensuration assessment in 1995 as a surrogate for total
biomass. The relationships for simulating wood (branch, stem and root) respiration are shown
in Table 7.
site
timber volume (1995; m3 ha-1) wood respiration (µ
µmol m-2 s-1)
(0.087(Ta-15))
Alice Holt
214
Rw=3.357exp
Grizedale
172
Rw=2.699exp(0.087(Ta-15))
Table 7. Timber volume measured in 1995 and expression describing single system wood respiration.
The relationships outlined in Table 7 include a contribution from foliage decomposition as
they were derived from flux measurements rather than process descriptions of the respiration
components from the individual organs. Since LAI and thus leaf biomass at Alice Holt was
considerably larger than that at Grizedale (6.4 and 2.5 tC ha-1 vs 3.08 and 1.25 tC ha-1), the
annual carbon fluxes for Grizedale given in Tables 9, 12&14 have been corrected for these
differences in CO2 emissions from leaf decomposition (Rl), amounting to 1.25 tC ha-1 for the
baseline simulation. Other parameter values used as input to the model are given in Table 8,
including both those observed or derived during the course of the project, and those extracted
from the scientific literature.
40
Parameter
unit
Alice Holt Grizedale source
maximum stomatal conductance, gsmax
mol m-2 s-1
0.4
0.4
observed
foliage water holding capacity, Ileaf
mm m-2
0.205
0.205
Harding et al., 1992
field capacity of soil, SMsat
m3 m-3
0.51
0.42
observed
available soil water, SMsat-SMmin
m3 m-3
0.42
0.26
observed
soil depth
m
1.0
0.5
observed
end of growing season
day
280
280
observed
light extinction coefficient, Kext
0.4
0.4
observed
maximum vpd for stomatal conductance
kPa
4.76
4.76
Broadmeadow et al., 1999
Jmax
132.6
124.6
observed
µmol m-2 s-1
Vmax
47.2
44.5
observed
µmol m-2 s-1
-2 -1
Rd
1.4
1.26
observed
µmol m s
SLA
cm2 g-1
102.3
89.9
observed
Table 8. Parameters used as input for the baseline annual carbon flux simulations given in Table 9.
The results for the baseline simulations given in Table 9 include estimates of interception (Ei)
and transpiration (Et) losses of water vapour, gross photosynthetic productivity (GPP), foliage
respiration (Rf), wood respiration (Rw), and net primary productivity (NPP; =GPP-Rf-Rw). An
indication of direct effects of climate on productivity is provided by the simulations in which
each site is subject to the climate of the other.
simulation
Alice Holt baseline
Grizedale baseline
GPP
tC ha-1 y-1
17.20
10.22
Rf
tC ha-1 y-1
4.44
2.18
Rw
tC ha-1 y-1
8.35
4.94
Ei
mm y-1
81.9
44.1
Et
mm y-1
267
93
NPP
tC ha-1 y-1
4.35
3.11
Alice Holt with Grizedale met.
14.56
3.83
7.73
64.6
125
3.00
Grizedale with Alice Holt met.
11.50
2.52
5.43
95.2
172
2.30
Table 9 Baseline simulations of carbon and water fluxes for 1999 at Alice Holt and Grizedale. Rw for the
Grizedale simulations has been corrected for Rl, the difference in leaf decomposition rates between the two sites.
The net ecosystem flux (equivalent to NPP) measured at Alice Holt by eddy correlation in
1999 was 3.88 tC ha-1. These measurements were, however, made in a block of woodland
which was thinned more recently than the intensive monitoring plot, and as a consequence,
LAI is lower (~4.5 vs 6.41), accounting for the difference in NPP. Simulated rainfall
interception is 9.3% at Alice Holt, a value much lower than that estimated from throughfall
collection (16%; 1995-1997). However, the model does not account for interception by stem
and branch wood, which accounts for 9.5% as indicated by winter (week 1-18) rainfall
interception (1996-1997), and thus the interception estimates modelled here are appropriate.
Transpiration losses are slightly lower than would be expected from traditional models of
forest water use (~300-350 mm; eg Harding et al., 1992), although simulated soil moisture
content is close to that measured during the course of the project by TDR (Fig. 27). The
model also does not assess evaporation from bare soil, or the evapo-transpiration component
of ground vegetation.
41
(a)
soil moisture content (% v/v)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
(b)
0
100
200
day of year
300
400
0
100
200
day of year
300
400
soil moisture content (% v/v)
0.6
0.5
0.4
0.3
0.2
0.1
0
Fig. 27. Comparison of measured and modelled soil moisture content at (a) Alice Holt (modelled: fine line 0-20
cm, bold line 20-50 cm; measured: solid squares 0-15 cm, crosses 15-30 cm, open squares 30-60 cm) and (b)
Grizedale (fine line: modelled 0-20 cm; solid line: measured 30 cm).
At Grizedale, the much lower incident radiation (2864 MJ vs 4081 MJ) accounts for the lower
GPP. Together with lower air temperature and higher vapour pressure, the low global
radiation also results in very low estimates for both interception and transpiration losses.
Throughfall data indicate total interception losses (including stem interception and stemflow)
of 24%. Estimated NPP (3.11 tC ha-1 y-1) is lower than that at Alice Holt, and is reflected in
their site indices (GYC6 at Alice Holt, GYC4 at Grizedale). The reciprocal simulations
indicate the relative impact of the different climatic conditions and stand structure on NPP
and water balance estimates. The Grizedale climate reduced GPP and transpiration losses at
Alice Holt by 15% and 53%, respectively. Furthermore, although annual rainfall at Grizedale
is approximately twice that of Alice Holt, foliage interception losses were reduced by 21%
when the Grizedale climate data were used for the Alice Holt stand. The Alice Holt climate
increased NPP by 25% at Grizedale, whilst transpiration losses were increased by 85%. The
baseline data also demonstrate that small changes in estimates of any of the carbon fluxes can
have a very large effect on estimates of NPP and thus site productivity, as NPP generally
represents only ~25% of GPP. It is therefore crucial that any variation introduced through
different methodologies of deriving meteorological inputs is assessed and minimised.
3.6.4 Sensitivity analysis of meteorological data
Simulations carried out using meteorological data from a within-plot AWS are compared
with simulations in which data from a daily climatological station are used to generate hourly
meteorological data, and also for data from an AWS adjacent to the climatological station
42
(open AWS; Table 10). In addition, the sensitivity of simulated NPP to each meteorological
variable is given for the baseline simulation using within-plot AWS data.
GPP
17.15
16.28
18.06
18.11
16.35
18.04
17.87
18.06
18.09
17.15
17.26
16.84
17.24
16.32
17.45
NPP
4.35
1.56
2.96
3.02
1.26
2.87
2.78
3.41
2.99
2.59
4.30
4.04
4.44
3.52
4.65
Rf
4.44
4.60
4.53
4.53
4.53
4.60
4.53
4.53
4.53
4.44
4.60
4.44
4.44
4.44
4.44
Rw
8.35
10.12
10.57
10.57
10.57
10.57
10.57
10.12
10.57
10.12
8.35
8.35
8.35
8.35
8.35
EI
81.9
79.1
108.3
85.8
107.1
107.1
105.1
108.3
104.7
81.9
82.5
85.0
77.2
80.7
81.9
Et
267
270
246
266
245
244
250
246
244
267
274
285
256
255
287
Scenario
baseline
open AWS met
manual met
manual met + open AWS rain
manual met + open AWS Q
manual met + open AWS Ta
manual met + open AWS RH
manual met + open AWS Ts
manual met + open AWS WS
canopy met + open AWS Ts
canopy met + open AWS Ta
canopy met + open AWS RH
canopy met + open AWS WS
canopy met + open AWS Q
canopy met + open soil moisture
16.77
17.38
17.15
17.15
17.02
17.23
16.28
17.76
17.43
17.45
17.23
17.08
16.23
17.44
14.75
18.81
4.55
3.93
5.69
2.76
4.22
4.44
3.48
4.96
4.64
4.65
4.43
4.29
3.44
4.64
1.95
6.01
3.86
5.10
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
4.44
8.35
8.35
7.02
9.95
8.35
8.35
8.35
8.35
8.35
8.35
8.35
8.35
8.35
8.35
8.35
8.35
79.5
84.1
81.9
81.9
78.2
84.8
80.3
83.3
81.9
81.9
78.3
84.8
116.0
60.5
81.9
81.9
250
282
267
267
261
271
249
283
286
288
258
274
309
206
278
252
- 2 degree Ta
+ 2 degree Ta
- 2 degree Ts
+ 2 degree Ts
- 20% rain
+ 20% rain
- 20% Q
+ 20% Q
- 10% SM
+ 10% SM
- 20% ws
+20% ws
- 20% vp
+ 20% vp
-100 ppm CO2
+100 ppm CO2
Table 10. Results from a sensitivity analysis of the effects of different meteorological inputs and manipulation of
those inputs on model output for annual carbon and water flux simulations at Alice Holt.
The input of open AWS meteorological data led to a dramatic reduction in NPP of 74% in
comparison to the baseline scenario where data from the within-plot AWS was used. The use
of daily climatological data reduced the difference, although a considerable effect was still
apparent (32% reduction in NPP). The substitution of individual variables into the within-plot
dataset suggest that Ts is an important driver of NPP. This results from the fact that Ts within
the plot is 1-2oC cooler than at the other two weather stations as a result of direct solar
radiation not reaching and therefore warming the soil (Table 11).
43
data source
Alice Holt
climatological
open AWS
within plot AWS
Ta (mean)
o
C
Ts (mean)
o
C
RH (mean)
%
ws (mean)
m s-1
Q (total)
MJ
R (total)
mm
11.4
10.9
10.5
11.9
11.0
9.7
83.9
32.1
85.1
1.24
1.17
2.44
3499
3486
4081
850
885
-
2720
2864
2021
1665
Grizedale
climatological
9.3
10.4
85.4
1.34
within plot AWS
9.3
8.8
88.1
1.56
Table 11. Summary of meteorological data for 1999 at Alice Holt and Grizedale.
However, soil temperature does not account for the entire difference, and it is apparent that
the solar radiation input from the open AWS results in very low estimates of GPP (and thus
NPP) compared to the within-plot AWS. The difference in radiation input between the two
AWSs accounts for this difference in GPP between the baseline scenario and the open AWS
scenario, and results from the open AWS being shaded by trees for a proportion of the day.
However, it does not account for the large difference in GPP and NPP between the manual
met. scenario and the manual met+open AWS Q scenario. Here, annual incident radiation is
almost identical (which is encouraging as the daily to hourly weather generator was calibrated
using 1995-1996 data from these two weather stations), and the resulting difference must be a
result of differences in the partitioning of radiation through the course of the day. This overriding effect of solar radiation highlights the importance of measuring radiation input at a
suitable time resolution, and also the importance of siting the sensor away from any influence
of shade.
Estimates of interception losses are ~20% higher when the climatological dataset is used as
input rather than the hourly open AWS data. Although rainfall was distributed through the
day as a function of intensity, this distribution does not mirror reality, and again highlights the
importance of measuring at a suitable timeframe. Lower humidity at the open AWS resulted
in a significant reduction in GPP and NPP compared with the baseline scenario. The lower
wind speed recorded at the climatological station resulted in lower interception and
transpiration losses and higher GPP and NPP reflecting water limitation of CO2 assimilation
during the summer. This is confirmed by the increase in GPP simulated by using soil moisture
at 30 cm as a direct input, rather than modelled moisture content. The sensitivity analysis
confirms the findings described above that of the meteorological variables assessed, soil
temperature and incident solar radiation have the most influence on NPP. In addition, the
sensitivity analysis also indicates that a 20% reduction in vapour pressure would limit
productivity, presumably as a result of stomatal closure at low humidity limiting CO2 uptake.
The response of the model output to variation in CO2 atmospheric concentration was included
in the simulations to demonstrate how a model of this type can be applied to climate change
scenarios. A dramatic effect on GPP and NPP is apparent, both as a result of a 100 ppm
reduction (to a pre-industrialisation concentration), and the 100 ppm increase (predicted by
the middle of this century (IPCC, 1995). However, it should be borne in mind that this
response to CO2 concentration (the CO2 fertilisation effect) does not account for changing
resource allocation or canopy structure, nor does it consider possible long-term nutrient
deficiencies that may occur if these enhanced growth rates are maintained (Eamus and Jarvis,
1989). The same is also true for long-term perturbations in other meteorological variables.
Similar results of the sensitivity analysis are seen in the Grizedale simulations (Table 12). The
impact of generating hourly rainfall and solar radiation data is again apparent, with large
44
differences observed in Ei and GPP, respectively. Soil temperature, solar radiation and
atmospheric CO2 concentration are again highlighted as having over-riding influence on
potential productivity. Grizedale is a more marginal site as a result of the lower solar radiation
input, and here, NPP is a smaller proportion of GPP. As a consequence small changes in the
carbon pools or fluxes have an even larger effect on NPP. This is likely to be the case for all
marginal forest sites.
GPP
NPP
Rf
Rl
Rw
Ei
Et
scenario
10.22
11.70
10.29
11.46
11.53
11.71
11.70
11.70
10.25
3.11
3.33
1.92
3.09
3.25
3.33
4.50
3.33
3.14
2.18
2.26
2.26
2.26
2.18
2.26
2.26
2.26
2.18
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
1.3
6.19
7.36
7.36
7.36
7.36
7.36
6.19
7.36
6.19
44.1
95.2
97.2
63.4
89.6
73.6
95.2
81.6
44.1
93
111
102
122
103
89
111
100
94
baseline
manual met
manual met + AWS Q
manual met + AWS rain
manual met + AWS Ta
manual met + AWS RH
manual met + AWS Ts
manual met + AWS WS
canopy met + soil moisture
9.81 2.98 1.89
1.3
6.19
42.0 83
- 2 degree Ta
10.54 3.11 2.50
1.3
6.19
46.3 102
+ 2 degree Ta
10.22 4.09 2.18
1.3
5.21
44.1 93
- 2 degree Ts
10.22 1.91 2.18
1.3
7.38
44.1 93
+ 2 degree Ts
10.22 3.11 2.18
1.3
6.19
43.4 93
- 20% rain
10.22 3.11 2.18
1.3
6.19
44.6 92
+ 20% rain
9.37 2.25 2.18
1.3
6.19
42.1 79
- 20% Q
10.77 3.66 2.18
1.3
6.19
45.9 105
+ 20% Q
10.22 3.11 2.18
1.3
6.19
41.5 90
- 20% ws
10.22 3.11 2.18
1.3
6.19
46.6 95
+20% ws
10.09 2.98 2.18
1.3
6.19
72.7 130
- 20% vp
10.24 3.13 2.18
1.3
6.19
31.3 69
+ 20% vp
8.93 1.82 2.18
1.3
6.19
44.1 97
-100 ppm CO2
11.04 3.92 2.18
1.3
6.19
44.1 87
+100 ppm CO2
Table 12 Sensitivity analysis of the effects of different sources of meteorological data, and the impact of specific
changes in those variables on annual carbon and water flux simulations for Grizedale in 1999. Rl is the correction
applied to the wood respiration component as a result of the lower leaf litter production at Grizedale in
comparison to Alice Holt (see section 3.6.3).
3.6.5 Sensitivity analysis of selected model parameters
A sensitivity analysis was also carried out on the impact of varying the parameters given in
Table 8, in addition to variation in developmental phases. In order to simplify the simulations,
and the output from them, a simplified baseline scenario has been adopted, and model outputs
compared to this. This scenario is based upon physiological parameters derived from the
routine foliar collection in 1999. It is therefore assumed that all foliage has the same values
for Vmax, Jmax SLA and Rd throughout the leafed period based upon [N]=2.31% and
SLA=102.3 cm2 g-1. These simulations are therefore unrealistic in physiological terms , but
are useful in assessing the sensitivity to each of the parameters in turn. Leaf area development
is assumed to be linear over the dates of development observed in section 3.2, maintaining
LAImax (from litter collection) until the designated end of the growing season. Each parameter
is varied by 20%, except for developmental inputs which are varied by ten days. It should also
be noted that for these simulations, no feedback loop between photosynthetic carbon
assimilation and growth or allocation is invoked, although this is achieved if the growth
module of GROMIT (or ForestGrowth) is implemented. Model outputs are therefore treated
in isolation – for example, a 20% increase in Vmax results in an approximate doubling of NPP,
but no change in respiration or evapo-transpiration outputs; in reality, the increase in Vmax is
likely to be a result of increased foliar [N]a , and thus respiration is likely to be higher, whilst
45
the increased NPP may result in increased leaf area together with larger evapo-transpiration
losses of water vapour. However, these simulations do provide a useful indication of the
sensitivity of the simulation outputs to each of the parameters in turn.
GPP
NPP
Rf
Rw
Ei
Et
scenario
19.25
4.07
6.82
8.35
75.7
276
‘simplified’ baseline
18.91
3.74
6.82
8.35
76.3
276
Jmax = 106
19.47
4.29
6.82
8.35
76.3
276
Jmax = 159
17.81
2.64
6.82
8.35
76.3
276
Vmax = 37.8
20.31
5.14
6.82
8.35
76.3
276
Vmax = 56.6
17.70
3.89
5.45
8.35
66.1
260
LAImax = 5.1
20.54
3.98
8.20
8.35
89.0
289
LAImax = 7.7
20.63
5.03
7.24
8.35
80.7
294
budburst = day 111
17.88
3.11
6.41
8.35
68.7
262
budburst = day 131
20.00
4.53
7.11
8.35
79.5
284
canopy development = 24 days
18.69
3.73
6.60
8.35
72.6
270
canopy development = 44 days
18.35
3.56
6.44
8.35
67.0
264
end of growing season = day270
20.00
4.46
7.19
8.35
78.0
287
end of growing season = day 290
19.36
4.18
6.82
8.35
75.7
279
SMsat = 41%
19.25
4.07
6.82
8.35
75.7
276
SMsat = 61%
18.75
3.57
6.82
8.35
75.7
254
SMsat-SMmin = 34%
19.61
4.44
6.82
8.35
75.7
295
SMsat-SMmin = 50%
19.26
4.08
6.82
8.35
68.1
281
Ileaf = 0.164
19.24
4.07
6.82
8.35
82.6
271
Ileaf = 0.246
20.67
5.50
6.82
8.35
74.8
273
Kext = 0.32
18.13
2.95
6.82
8.35
76.3
279
Kext = 0.48
19.16
3.99
6.82
8.35
75.7
269
gsmax = 0.32
19.30
4.13
6.82
8.35
75.7
281
gsmax = 0.48
Table 13. Simulated response of model outputs to varying parameter values by + / – 20% for Alice Holt. The
baseline simulation is simplified in comparison to that given in Tables 9&10 (see text).
GPP and NPP were more sensitive to manipulation of Vmax than Jmax, with Vmax having a
particularly marked effect. However, both parameters did have a large effect on the estimates
of NPP. The response of NPP to LAI was minimal, with reductions in NPP predicted for both
a 20% increase and 20% decrease in leaf area. This indicates that 6.4 is close to optimum
canopy density as a result of the balance between photosynthesis and respiration, and suggests
that further increases in LAI are unlikely. However, this response belies the fact that varying
LAI did have a large effect on both GPP and foliage respiration losses. The model predicts
that all inputs relating to growing season longevity and development are critical, particularly
during budburst when radiation inputs are higher than at the end of the season. In this context,
the rate of canopy development is also important.
GPP NPP Rf
Rl Rw Ei
Et scenario
10.22 3.11 2.18 1.3 6.19 44.1 93 baseline
8.03 1.10 1.74 1.0 6.19 35.2 73 LAI = 2.46
11.16 3.87 2.61 1.5 6.19 48.6 100 LAI = 3.70
11.04 3.77 2.33 1.3 6.19 47.2 101 budburst = day 106
9.45 2.51 2.01 1.3 6.19 41.1 86 budburst = day 126
9.85 2.87 2.04 1.3 6.19 42.3 92 end of growing season = day 270
10.54 3.28 2.33 1.3 6.19 45.7 94 end of growing season =day 290
Table 14. Effect of canopy development inputs on model outputs for Grizedale in 1999.
A further demonstration of the impact of development and growing season longevity is given
in Table 14, where these specific simulation results are shown for Grizedale, and the more
marginal nature of the site magnify the effects of growing season length and canopy
development on NPP. Field capacity had little effect on model output, but a 20% reduction in
46
available water content did reduce NPP by 12%, and if all soil water was available, NPP
increased by 10%, indicating that NPP was moisture limited to a small degree in the baseline
scenario. The effect of foliage water holding (Ileaf) capacity on NPP was minimal although as
expected, a significant effect on Ei was apparent. The model output was relatively insensitive
to stomatal conductance in the absence of limiting factors (gsmax), which is fortunate, given the
difficulty of deriving an accurate value for this parameter. The influence of the effective light
extinction coefficient on GPP and NPP is overriding, and it is also a primary driver for
determining optimum LAI.
47
4 Conclusions
The sensitivity analyses outlined in this, and preceding sections thus highlight those
parameters and variables which act as the main drivers for predicting NPP, and thus for
modelling the effects of climate and environment on tree health and growth. Observations of
canopy development and maximum LAI are essential inputs, and any measurement of LAI
will also relate to the annual measurements of crown condition, providing a link between the
routine extensive measurements at Level I and II, and detailed modelling exercises of the type
outlined in this report. The determination of LAI and canopy light transmission will also
allow the derivation of an appropriate value for Kext, which is an essential input to the model,
and will vary with canopy structure. Foliar sampling on a more frequent basis will provide
additional information for model parameterisation. However, a more detailed analysis of
foliar chemistry at a single time-point after complete canopy development would be more
practical, and also improve the accuracy of model outputs to a greater extent. These additional
measurements cannot be made in isolation, and require the implementation of programmes of
work for model calibration and validation as outlined in other sections of this report.
Meteorological inputs to any growth model are an essential element to outputs representative
of a particular site. The analysis of the impact of different data sources has indicated that not
only must the data be representative of the site in the long term, but the time-frame of the
observations are also critical, particularly rainfall and solar radiation. Whilst acceptable
accuracy of model outputs can be obtained through using daily observations some distance
from the forest plot, others require sub-daily measurements, ideally at the forest plot.
Although the results from the sensitivity analysis and other model simulations do provide
some robust recommendations, any additional recommendations that are implemented should
fit within the remit of a long-term network of monitoring stations, and also be financially and
practically feasible. Furthermore, it is crucial that the work described here, is placed in the
context of the modelling approach that has been described. This approach is intentionally at
the intensive end of the modelling spectrum, and has been adopted to enable both the
assessment of the impacts of ozone on forest health and productivity, and the prediction of
future changes in forest growth as a result of global climate change and rising atmospheric
carbon dioxide concentrations.
48
5 Recommendations
EU Council Regulation 3528/86 for the protection of the forests of the European Community
requires that a monitoring programme is carried out to ensure the protection of the
Community’s forests. To date, this monitoring programme has concentrated on the impacts of
acid deposition, together with a number of projects which have focussed on developing the
intensive forest health monitoring programme to encompass a wider range of both natural and
anthropogenic drivers of forest health within the EU. To further this drive for the broadening
of the remit of the intensive forest health monitoring programme, the capacity to undertake
process modelling of forest growth and health should be made possible through a
development of the monitoring protocol defined within Council Regulation 1091/94. If this
strategy is adopted, it will further the programme’s ability to identify and predict the effects of
ozone pollution, rising atmospheric carbon dioxide concentrations, global climate change, and
interactions between these anthropogenic influences on vitality and yield of the community’s
forests, including the potential for long-term carbon sequestration. Furthermore, it will enable
more precise estimates of pollutant load to forest ecosystems to be made, and further develop
the capacity to derive cause-effect relationships.
The recommendations made here are intended to produce the most benefit to the programme
through a minimum resource input, using the ethos of a single protocol to be implemented by
all member states to produce a regional dataset that is applicable across the entire
Community. This approach can thus build upon the existing monitoring programme and
provide added value through its ability to inform on the wider debate of forest ecosystem
function. The specific recommendations are as outlined below:
• Each member state should select a number of forest plots across a range of potential
pollutant impacts and for sensitive species that should be developed as ‘level III’
monitoring plots, with the explicit aim of developing a protocol, including the derivation of
input parameters and variables, to enable process modelling of forest function and growth.
Careful consideration must be given within a Europe-wide context, to both the choice of
species and distribution of plots that are represented in any upgraded network. These plots
should be chosen in a decision making process that encompasses the recommendations of
other manuals and sub-manuals, to provide a limited number of plots which can provide a
bridge between the level II network and the handful of forest plots within Europe at which
a plethora of environmental, physical and physiological parameters and variables are
available. This decision making process should specifically include ambient air pollution
monitoring and within plot meteorology.
At each forest plot, the following measurements should be made as a minimum. It should
however be borne in mind that these recommendations refer specifically to broadleaf
deciduous species. Modifications may be required for coniferous and/or evergreen species.
• monitor litter fall to provide a measure of leaf area index; this approach will corroborate
the assessment of crown condition, and is also applicable to evergreen species, providing
information on inter-annual variation in needle retention. Further information will also be
provided on the relative resource allocation to flowering and fruiting, which can be
considerable and limit carbohydrate available for growth.
• monitor canopy development through measuring canopy light transmission on a continuous
basis.
• together with the light transmission measurements recommended above, a limited within
plot weather station should be established, including the measurement of the following
variables:
49
soil temperature
soil moisture in the predominant organic/surface and mineral/subsoil horizons
as a minimum
air temperature at crown height
vapour pressure if representative data from an AWS are not available in the
immediate vicinity (within ~10 km)
• the meteorological mast should be accessible to enable the routine monitoring of ambient
air pollution concentrations
• foliar nitrogen content should be measured on an annual basis, as a minimum. Specific leaf
area should be monitored on a more routine basis, preferably once per month
Plots should only be selected where long term meteorological records are available in the
immediate vicinity such that past and future changes in forest growth and condition can be
analysed against climatic variables.
The recommendations made here are by no means exhaustive, and further analysis and
assessment should be encouraged. However, these recommendations, when carried out in an
integrated programme alongside the full suite of measurements currently recommended under
the existing programme, will provide the minimum input variables and parameters to enable
process modelling of forest function particularly when applied to identifying the effects of
ozone pollution and global climate change.
Acknowledgements
This work was jointly funded by the EU under the UN/ECE ICP Forests Intensive Forest
Health Monitoring Programme and the UK Forestry Commission Policy and Practice
Division. The authors are grateful for the support provided by Paul Gough, Lenny Thornton,
and Ian Yoxall from the Kielder Technical Support Unit who carried out the regular site visits
at Grizedale with efficiency, to David Gregory and Tim Belino from the Forest Enterprise
Lakes District Office who provided on site support carried out the spring phenological
assessment, and to Ernest Ward, Chris Whitfield, Jo Edgerton Sue Benham and the team in
the Chemical Analysis Laboratory at Alice Holt for performing the foliar chemical analysis
and liasing over site visits.
50
References
Aubinet M, Grelle A, Ibrom A, Rannik Ü, Moncrieff J, Foken T,
Kowalski AS, Martin PH, Berbigier P, Bernhofer Ch, Clement R, Elbers J, Granier A,
Grünwald T, Morgenstern K, Pilegaard K, Rebmann C, Snijders W, Valentini R, Vesala T
(1999) Estimates of the annual net carbon and water exchange of European forests: the
EUROFLUX methodology. Advances in Ecological Research
Broadmeadow MSJ (1998) Ozone and Forest Trees. New Phytologist 139:123-125.
Broadmeadow MSJ, Heath J, Randle, TJ (1999) Environmental limitations to physiologically
effective O3 exposure. Water, Air and Soil Pollution 116:299-310.
Comins HN, McMurtrie RE (1993) Long-term response of nutrient-limited forests to CO2enrichment; equilibrium behaviour of plant-soil models. Ecological Applications 3: 666-681.
ECN (1996) The United Kingdom Environmental Change Network: protocols for standard
measurements at terrestrial sites. eds. JM Sykes, AMJ Lane. The Stationary Office, London.
ECOCRAFT (1999) Predicted impacts of rising carbon dioxide and temperature on forests in
Europe at stand scale. Final project report (ENV4-CT95-0077).
Edwards PN, Christie JM (1981). Yield models for forest management. Forestry Commission
Booklet Number 48. Forestry Commission, Edinburgh.
Evans SP (1995) TRESTLE – Tree response establishment and succession in the transient
landscape of Europe. Final report for European Commission contract EV5V-CT-94-5234.
Cranfield University, UK.
Friedli H, Lotscher H, Oeschger H, Siegenthaler U and Stauffer B (1989). Ice core record of
the 13C/12C ratio of atmospheric CO2 in the past two centuries. Nature 324:237-238.
Friend AD (1995) PGEN: an integrated model of leaf photosynthesis, transpiration, and
conductance. Ecological Modelling 77:233-255.
Friend AD, Stevens AK, Knox RG, Cannell MGR (1997) A process based, terrestrial
biosphere model of ecosystem dynamics (HYBRID v3.0). Ecological Modelling 95:249-287.
Gates DM (1980) Biophysical Ecology. Springer-Verlag, New York.
Goudriaan J (1988) The bare bones of leaf angle distribution in radiation models for canopy
photosynthesis and energy exchange. Agricultural and Forest Meteorology 43:155-169.
Hall DGM, Reeve MJ, Thomasson AJ, Wright VF (1977) Water retention, porosity and
density of field soils. Technical Monograph No 9. Soil Survey, Harpenden, UK.
Hanninen H (1990) Modelling bud dormancy release in trees from cool and temperate
regions. Acta Forestalia Fennica.
Harding RJ, Hall RL, Neal C, Roberts JM, Rosier PTW, Kinniburgh DG (1992) Hydrological
impacts of broadleaf woodland: implications for water use and water quality. UK DoE/NRA
Contract report 115/03/ST. National Rivers Authority, Bristol, UK.
51
IPCC (1995) Climate change 1994: radiative forcing of climate change and an evaluation of the
IPCC IS92 emission scenarios. Houghton JT, Meira Filho LG, Bruce J, Lee H, Callander BA,
Haites E, Harris N, Maskell K, eds. Cambridge, UK: Cambridge University Press.
Jarvis PG (1976) The interpretation of the variations in leaf water potential and stomatal
conductance found in canopies in the field. Philosophical Transactions of the Royal Society of
London, series B 273: 593-610.
Jarvis PG (1993) Maestro: a model of CO2 and water vapour exchange by forests in a globally
changed environment. In: Schulze D.M., Mooney H.A. (eds.): Design and execution of
experiments on CO2-enrichment. Cec, Brussels, pp. 107-116.
Kellomäki S, Väisänen H, Strandman H (1993) FinnFor: a model for calculating the response
of boreal forest ecosystem to climate change. University of Joensuu, Faculty of Forestry,
Research Note 6:1-120.
Kull O, Jarvis PG (1995) The role of nitrogen in a simple scheme to scale up photosynthesis
from leaf to canopy. Plant, Cell and Environment 18:1174-1182.
Kull O, Kruijt B (1999) Acclimation of photosynthesis to light – a mechanistic approach.
Functional Ecology 13:
Kull O, Kruijt B, Broadmeadow M, Meir P (1999) Light distribution and foliage structure in an
oak canopy. Trees 14:55-64.
Kull O, Niinemets Ü (1998) Distribution of leaf photosynthetic properties in tree canopies:
comparison of species with different shade tolerance. Functional Ecology 12:472-479.
McMurtrie RE, WANG Y-P (1993) Mathematical models of the photosynthetic response of
tree stands to rising CO2 concentrations. Plant Cell and Environment 16:1-13.
McMurtrie RE, Rook DA, Kelliher FM (1990) Modelling the yield of Pinus radiata on a site
limited by water and nitrogen. Forest Ecology and Management 30, 381-418.
Medlyn BE, Badeck FW, de Pury DGG, Barton CVM, Broadmeadow M, Ceulemans R, de
Angelis P, Forstreuter M, Jach ME, Kellomäki S, Laitat E, Marek M, Philippot S, Rey A,
Strassenmeyer J, Laitinen K, Liozon R, Poiter B, Roberntz P, Wang K, Jarvis PG (2000)
Effects of elevated [CO2] on photosynthesis in European forest species: a meta-analysis of
model parameters. Plant, Cell and Environment 22:1475-1495.
Meng FR, Arp PA (1994) Modelling photosynthetic responses of a spruce canopy to SO2
exposure. Forest Ecology and Management 67:69-85.
Monteith JL (1965) Evaporation and Environment. In ‘The state and movement of water in
living organisms’. Symposium of the Society for Experimental Biology 19:205-234.
Nikolov NT, Zeller KF (1992) A solar radiation algorithm for ecosystem dynamics models.
Ecological Modelling 61:149-168.
Pyatt DG (1982) Soil classification. Forestry Commission Research Information Note
68/82/SSN. Forestry Commission, Edinburgh.
52
Randle TJ, Ludlow AR (1998) Growth costs to plantations: modelling the cumulative effects
of defoliation by Elatobium abietinum. In: The green spruce aphid in western Europe:
Ecology, status, impacts and prospects for management. Forestry Commission Technical
Paper 24. Forestry Commission, Edinburgh, pp. 32-40.
Sanz MJ, Millan MM (1998) The dynamics of polluted airmasses and ozone cycles in the
western Mediterranean: relevance of forest ecosystems. Chemosphere 36:1089-1094.
Sonntag M (1997) Comparison of forest models at the leaf and canopy scale. In Impacts of
Global Change on Tree Physiology and Forest Ecosystems (eds. G.M.J. Mohren, K. Kramer
and S. Sabaté), pp. 15-25. Kluwer Academic Publishers, Dordrecht.
Taylor CMA (1991) Forest fertilisation in Britain. Forestry Commission Bulletin 95. Forestry
Commission, Edinburgh.
Ulrich B (1983) Soil acidity and its relation to acid deposition. In: Effects of Accumulation of
Air Pollutants in Forest Ecosystems (eds. B. Ulrich, J. Pankrath). pp 127-146. D. Reidel,
Boston.
UN-ECE. 1996. Karenlampi L, Skarby L, eds, Critical Levels for ozone in Europe: testing and
finalizing the concepts, Workshop Report. Kuopio, Finland. 72-82.
UN/ECE and EC (1997) Ten Years of Monitoring Forest Condition in Europe – technical
background report. eds. C. Muller-Edzards, W. De Vries, JW Erisman. UN/ECE, Brussels.
Väisänen H, Strandman H, Kellomäki S (1994) A model for simulating the effects of
changing climate on the functioning and structure of the boreal forest ecosystem: an approach
based on object-oriented design. Tree Physiology 14:1081-1095.
von Caemmerer S, Farquhar GD (1981) Some relationships between the biochemistry of
photosynthesis and the gas exchange of leaves. Planta 153, 376-387.
Wang Y-P, Jarvis PG (1990) Description and validation of an array model - MAESTRO.
Agricultural and Forest Meteorology 51:257-180.
Whitehead D, Kelliher FM (1991) A canopy water balance model for a Pinus radiata stand
before and after thinning. Agricultural and Forest Meteorology 55, 109-126.
Appendices
I Model description and Validation of ForestFlux (GROMIT)
II Reporting forms
III Pictorial representation of standard flushing stages
IV Glossary of abbreviations and definition of symbols
53
Appendix I: Technical manual for ForestFlux (GROMIT)
1.0 Overview
2.0 Assimilation model
2.1 meteorology
2.2 canopy structure
2.3 light interception
2.3.1 Beam and diffuse light
2.3.2 sunlit and shaded foliage light intensities
2.3.3 calculation of sunlit and shaded foliage proportions
2.4 photosynthesis
2.5 stomatal conductance
2.6 water-balance
2.6.1 diffusion within the soil
2.6.2 interception and throughfall
2.6.3 evapo-transpiration
2.7 Respiration Overview
2.8 Calculation of growing season
2.8.1 bud-burst
2.8.2 seasonal development
2.8.3 End of season
2.9 Initialisation
3.0 Model validation
References
54
1.0 Overview
The modelling of long-term forest growth is achieved by coupling the annual output from a subdaily canopy assimilation module written in FORTRAN (ForestFlux) to a growth module that uses
the simulation language DARE-P (Lucas, 1974; Wait and Clarke, 1978). The modules are
fundamentally process based, although there are some elements of empirical sub-models. The
advantage of process based models, is that they can provide predictions of growth where
extrapolation is necessary – a failing largely inherent with empirical models.
The model is not structured as a single formulation, but is tailored to specific requirements,
including the use of alternative sub-models. ‘Add-ons’ have been developed for specific purposes.
In this description, only brief details of such add-ons are given, but where a function is integral to
the model, it is described more fully.
The objective of the canopy assimilation module (ForestFlux) is to calculate carbon uptake for a
given leaf area index (LAI). The well characterised biochemical formulation of photosynthesis
developed by Farquhar and co-workers (Farquhar and von Caemmerer, 1982; von Caemmerer et al.,
1994) are used, together with a stomatal conductance model after Jarvis (1976). Dates representing
bud-burst and end of season (senescence in broadleaf trees) are also modelled. This module has
been used for some basic flux simulations, although it does not make any claims to be a full flux
model as it includes no representation of canopy CO2 storage. Its primary objective, within the
framework of the growth model, is to define the relationship between LAI and assimilated carbon.
The time-steps of calculation are necessarily short (usually hourly), although once basic
relationships for a given site have been defined, then this intensive modelling approach may not be
necessary.
The growth module uses the relationship derived between LAI and assimilated carbon, and allocates
the carbon to compartments within the trees. The compartments consist of foliage, branches, stem,
coarse (transport) roots and fine roots. Allocation between compartments follows the pipe model
theory of Shinozaki et al. (1964). Each tree has a physical X-Y co-ordinate, and the space available
to the tree is based on a polygon taking account of the size and location of its neighbours. The
growth module uses annual time steps at present, although there is potential to simulate shorter
periods.
The description of the growth module is not given in this Appendix, but has been included in this
overview to indicate how long-term forest growth can be simulated using the annual process outputs
that have been described in the main report.
55
2.0 Description of assimilation module
2.1 Basic inputs: Meteorology
The simulation of carbon balance begins with the input of meteorological data. Typically, hourly
data are used with inputs of solar radiation, air temperature, humidity (or wet-bulb temperature),
wind-speed and precipitation, together with ambient carbon dioxide concentration. Such detailed
datasets are often not available, especially for historical records and future predictions. In these
cases, hourly data are derived from daily data using the following routines:
Temperature and Radiation. Where daily values of temperature and radiation are available, sine
curves are fitted with peaks at 14:00 hours and 12:00 hours respectively. A methodology has been
developed to allow the derivation of hourly radiation estimates from historical data recorded as
sunshine duration (Randle, 1997).
Relative Humidity. If humidity is not available, it is calculated from wet-bulb and dry-bulb
temperatures. Where daily values are given, it is assumed that vapour pressure remains constant
throughout the day and relative humidity changes with temperature. For temperatures above
freezing, the saturated vapour pressure ( Svp ) is given by:
Svp = 6.1708 exp (17.269 Tk ) /( 237.3+Tk )
2.1
where, Tk is the temperature in kelvin, and the vapour pressure ( Vp ) is
Vp = Svp wet − 0.66 (Tdry − Twet )
2.2
with Tdry and Twet , the dry-bulb and wet-bulb temperatures.
Relative humidity, Rh, is calculated from the vapour pressure and the saturated vapour pressure (at
Tdry; VPdry) at the relevant time:
Rh =
Vp
100
Vp dry
2.3
Windspeed and Precipitation. If hourly data are not available, a function must be developed
which describes hour to hour variation. This function will be site specific.
Ambient Carbon dioxide Concentration. If measured data are not available a simple function is
used, with maximum and minimum concentrations at dawn and two hours after dawn, respectively.
The concentration decreases linearly between the maximum and minimum values, remains at the
minimum value until two hours prior to dusk, and then increases linearly up to the maximum value
at dawn.
2.2 Basic inputs 2: Canopy structure
The most appropriate way to envisage how the model considers the structure of the canopy of a tree
is as a series of concentric shells (an analogy is a ‘Russian doll’). For ease of reference, the term
‘zone’ is used to represent each shell. Each ‘zone’ has a number of layers of foliage (definable), and
each layer has a proportion of foliage that is either sun-lit or shaded. The physiological
characteristics (Rd, Vmax, Jmax, gsmax etc.) are constant within each ‘zone’, but may vary between
‘zones’. To this end, there are many input and parameters that are required to define the canopy.
56
2.3 Light interception
Previously published formulations of the model have simulated CO2 assimilation as light
interception using the Beer-Lambert Law (Randle and Ludlow, 1998; Ludlow et al., 1990). The
model now incorporates a biochemical photosynthesis routine, based on the model of Farquhar and
co-workers (see Medlyn et al., 1999).
The canopy is divided into several ‘zones’ and each ‘zone’ has a maximum photosynthetic capacity
explicitly defined in the parameterisation. Photosynthetically active radiation (PAR) falling on each
layer of foliage is attenuated by the canopy leaf area extinction coefficient, such that the total PAR
absorbed is given by
I abs = I 0 (1 − exp( − Kext LAI ))
2.4
where, I abs is absorbed PAR, I 0 the PAR above the canopy, and Kext the light extinction
coefficient of the canopy.
ForestFlux treats light interception by shaded and sunlit foliage separately (Spitters, 1986b) and
integrates these processes with a biochemical CO2 assimilation model following Meng and Arp
(1994). Their approach has been further developed, through assigning each ‘zone’ specific
physiological characteristics, with the total number of zones limited to twelve. Each zone has a
defined number of foliage layers (leaf area index, LAI), with each layer of foliage (irrespective of
the number of layers in a ‘zone’) considered in turn, and calculated incident PAR falling on shaded
and sunlit leaves used to drive the biochemical model. If a layer spans two ‘zones’, weighted
averages of parameters are input into the biochemical model.
2.3.1 Partitioning of diffuse and direct light above the canopy
In order to treat sun-lit and shaded foliage separately, it is necessary to partition incoming radiation
into diffuse and direct light.
The amount of cloud cover has a large effect on the proportion of direct and diffuse radiation. All
irradiation is diffuse under an overcast sky but 23% is diffuse (with 77% of global irradiation direct)
under a clear sky, (Spitters, 1986a). Historical meteorological records have consisted of daily
measurements of sunshine duration, which have been used to derive the diffuse and direct
components of PAR using functions dependent on zenith angle, atmospheric pressure and potential
solar radiation. (Weiss and Norman, 1985). More recently, following the development of electronic
sensors and data logging capabilities, PAR or global radiation is measured directly. These data, at a
sub-daily timestep are used in many carbon flux and process-based models of forest growth such as
GROMIT (Ludlow et al., 1990), MAESTRO (Wang and Jarvis, 1990) and Biomass (McMurtrie et
al., 1990).
In order to provide these estimates of direct and diffuse radiation input, ForestFlux uses the
approach of Spitters (1986a). In this approach, the total extra-terrestrial radiation at a plane parallel
to the surface of the earth is given by:
S 0 = S c (1 + 0.033 cos(360 d/365))cosθ z
2.5
Where d is the day of year, Sc is the solar constant (1370 W/m2), and θ z is the azimuth angle of the
sun.
A four-part function describes the ratio of diffuse radiation (Sdf) to total radiation at the earth
surface(Sg), as a function of radiation at the earth surface and the extra-terrestrial radiance (So) (de
Jong, 1980). Both daily and hourly values can be derived (Spitters, 1986a; de Jong, 1980). The
relationships are notably constant over a range of climates and latitudes (Collares-Pereira and Rabl,
1979; Erbs et al., 1982) and are appropriate to a wide range of conditions. Since ForestFlux usually
calculates photosynthesis at an hourly time-step, the most appropriate function is:
57
S df /S g
S df /S g
S df /S g
S df /S g
=1
= 1 − 6.4 (Sg /S 0 − 0.22) 2
= 1.47 − 1.66 (Sg /S 0 )
=R
for
for
for
for
S g /S 0 ≤ 0.22
0.22 < S g /S 0 ≤ 0.35
0.35 < S g /S 0 ≤ K
K < S g /S 0
2.6
where R = 0.847 − 161
. cosθz + 104
. cos2 θz and K = (1.47 − R) / 1.66 .
2.3.2 Calculation of sunlit and shade light intensities
Only photosynthetically active radiation (PAR) is considered, and it is assumed that reflection is
negligible, although scattering of direct light is taken into account. The separation of PAR into
diffuse and direct components is explained in the previous section. Following Neumann et al.
(1989) and Meng and Arp (1994), PAR on sunlit leaves in the ith layer is calculated as:
PAR sun (i ) = PAR dir 0 (cos Ψorient / sin θ z ) + PAR shade (i )
2.7
where, PARdir0 is the amount of direct PAR above the canopy, Ψorient is the orientation angle of the
leaf in relation to the sun, and θz is the azimuth angle. In most circumstances, PARsun will be less
than the incoming radiation as the surface of the canopy is not a plane surface.
The calculation of PAR incident shaded leaves is more complex, as they receive diffuse light,
attenuated direct light and scattered direct light. Since each layer of leaves is treated separately, the
average light within the layer is not calculated, although incident light above each layer is a
necessary intermediate; for shaded leaves
PAR shade = PAR diff 0 exp( − K df
PAR dir 0 {exp[ − (1 − σ c ) 0.5 K bl
exp( − K bl
LAI(i ) − 1) +
LAI(i ) − 1] −
2.8
LAI(i ) − 1)}
where PARdiff0 is diffuse PAR above the canopy, Kdf the canopy extinction coefficient for diffuse
light, and σc is the scattering coefficient of single leaves. Kbl is the black body canopy extinction
coefficient (Spitters, 1986b, Meng and Arp, 1994) such that
K bl =
K df
1.6(1 − σ c ) 0.5 sin θ z
2.9
If light extinction is constant throughout the canopy, then equations (2.7) and (2.8) apply. In
circumstances where extinction is not constant, an approximation is made, where for sequential
layers of leaves, Kdf becomes an average of the layers above, and Kbl is calculated (equation 2.9)
using this value of Kdf.
2.3.3 Calculation of sunlit foliage
In any layer, the foliage is either sunlit or shaded. The proportion of sunlit foliage in each layer is
variable, and based on a Markov gap-frequency model, following Neumann et al., (1989) and Meng
and Arp (1994). A clumped foliage distribution is assumed, with leaf distribution in one layer
dependent on the distribution in previous layers. For each layer of foliage, the proportion which is
sunlit in layer, i, is defined as
58
Psun
i
æ
ö
ç − Ω 0 G LAI(n)
n =1
= expç
ç
cosθ z
ç
è
2.10
Where G is the leaf orientation function and Ω0 is the degree of leaf dispersal. For the first layer
(top of canopy), i=1, and i=2,3, etc., for subsequent layers.
2.4 Photosynthetic Assimilation
The evaluation of incident PAR on shaded and sunlit foliage in each layer enables the calculation of
carbon assimilation. The photosynthetic rate is limited by either the regeneration of ribulose 1,5biphosphate (RuBP), (Ac) or by electron transport (Aj). The formulation follows von Caemmerer et
al. (1994) and Medlyn et al. (1999).
A net = min(A j , A c ) − R d
2.11
Where R d is the dark respiration. A c and A j are given as:
Ac =
Vmax (Ci − Γ * )
C i + Km
Aj =
and
J ( Ci − Γ * )
4( C i + 2 Γ * )
2.12
Km is the Michalis-Menton coefficient for the reaction (Km = Kc (1+Oi / Ko)), with OI, the internal
oxygen concentration; Kc and Ko temperature dependent Arrhenius (exponential) functions:
æ 59400(T − 298) ö
K c = 404 expçç
è (298 R T )
and
æ 36000(T − 298) ö
÷÷
K o = 248 expçç
è (298 R T )
2.13
where R is the universal gas constant.
Γ∗ is the CO2 compensation point, a function of temperature (in kelvin):
Γ * = 36.9 + 1.88(T − 298) + 0.036(T − 298) 2
2.14
and J is a function of light, light absorption and quantum yield (α) such that:
θ J 2 − ( Iα + J max ) J + Iα J max = 0
2.15
where internal CO2 concentration (Ci), is a function of ambient CO2 concentration (Ca), net
assimilation (Anet) and stomatal conductance to water vapour (gs) such that
Ci =
Ca − Anet
16
. gs
2.16
The values Vmax and Rd are temperature dependent Arrhenius (exponential) functions:
Arrhenius:
æ H a (T − Tref ) ö
expç
ç T RT
ref
è
Jmax may take the form of either an Arrhenius function, or a Lloyd (humped) function:
59
2.17
æ D T − Hd
æ H a (T − Tref ) ö æ
÷ ç1 + expç s ref
expç
ç
ç T RT ÷ç
R Tref
ref
è
è
è
æ
æ D T − H d öö
ç1 + expçç s
÷÷
ç
RT
è
è
Lloyd:
öö
÷÷
÷÷
2.18
The Arrhenius function requires an input parameter of the activation energy of the relevant process,
Ha, together with Temperature, T (in kelvin), and the reference temperature for the value, Tref, (also
in kelvin). In addition to the parameters of the Arrhenius function, the Lloyd function requires
additional parameter values for the de-activation energy (Hd), and an entropy term (Ds).
Calculations are performed layer by layer for both the sunlit and shaded foliage. The product of Anet
and the proportion of foliage that is either sunlit or shaded gives carbon assimilation in that layer.
The summation of successive layers gives the cumulative assimilation for an increasing number of
layers of foliage.
2.5 Stomatal conductance
Calculation of stomatal conductance is made on a layer by layer basis following Jarvis (1976).
Response functions to environmental variables (varying between 0 and 1), are imposed on the
maximum value for stomatal conductance (gsmax), which is variable between ‘zones’.
g s = g s max f (Q) f (T ) f (ψ ) f (ϕ ) f (C )
2.19
where f (x) is the response function to light, temperature, humidity, soil-moisture and ambient
carbon dioxide concentration.
The light response function is defined as:
f (Q) =
S 0 (Q + q )
g s max + S 0 (Q + q )
2.20
where S0 is the slope of the light response function at low light levels, Q is incident PAR and
q=gdark/gsmax, with gdark defined as stomatal conductance in the dark.
The temperature response function is defined as:
f (T ) =
T − Topt æ Tmax − T ö
ç
Topt − Tmin çè Tmax − Topt
æ Tmax −Topt ö
ç
ç Topt −Tmin
è
2.21
where T (οC) is temperature, and Tmax, Topt and, Tmin are the maximum, optimum and minimum
temperatures for stomatal conductance.
The response function to vapour pressure deficit (humidity function) is defined as:
ì
0,
ï
ï
ï V pd − V
f (ψ ) = í1 −
,
V
V
−
2
1
ï
ï
1,
ï
î
ü
ï
ï
ï
≥ V1 ý
V pd > V2
V2 ≥ V pd
V1 > V pd
60
2.22
where V1 and V2 are the values of vapour pressure deficit at which gsmax is at a maximum and
minimum, respectively, and Vpd = (1-(Rh/100))V
The response to soil-moisture deficit (Smd) is defined as:
f (ϕ ) = max(1 − exp Sm1 ( S md − S m 2 ) ,0.0)
2.23
where Sm1 is a constant and Sm2 is the soil moisture deficit at which stomatal closure is complete.
The ground is assumed to be frozen (and thus f(ϕ)=0) when the air temperature is less then –1οC.
The response of stomata to ambient carbon dioxide concentrations is not well defined in literature.
However, an empirical function has been derived for oak (Broadmeadow et al., 1999):
f (C ) = max((1 − C1 (C a − 350),0.2)
2.24
where Ca is ambient carbon dioxide concentration, and Cl, is a response function parameter.
2.6 Water Balance
ForestFlux assumes three water holding zones in the soil. The number of zones can be extended, but
in practice three have been sufficient. Each of the zones is defined by a physical depth (which thus
gives rise to a volume under the projected area of the crown), and a saturated (holding) capacity.
The three zones are defined as:
Rooting zone: This is the only layer from which the tree draws water. Water movement by diffusion
from or to storage zone 1 is dependent on the difference in water content (or potential) between the
two zones.
Storage zone 1: water may flow by diffusion from or to the rooting zone as well as from and to
storage zone 2.
Storage zone 2: water may only flow between this zone and the storage zone 1.
In reality, the flow (diffusion) of water between zones will be continuous. However, for ease of
implementation, a time-step (or iterative) approach is used. The time step for evaluation (eg 1 hour)
is divided into 12 smaller units, termed mini-steps, and calculations of water-balance (transpiration,
precipitation and diffusion) are made over these shorter time-steps. This is repeated for each of the
remaining mini-steps, thus simulating equilibrium and reducing oscillations.
2.6.1 Diffusion
The transfer of water (Wt), from one zone into another, in time, t, (mini-step) is based upon the
difference in the water content between the zones(Wc1 – Wc2) and a diffusion constant, Dc:
Wt = t (Wc1 − Wc 2 ) Dc
2.25
2.6.2 Rainfall
The proportion of precipitation falling as throughfall depends on the quantity of water intercepted
and held within the canopy (leaf-water). If the water holding capacity of a layer is exceeded, then
subsequent rainfall is intercepted by the next layer. When all foliage layers in the canopy are at
capacity, additional precipitation falling within the time step is defined as throughfall.
The throughfall initially enters the rooting zone. The capacity of this zone depends on the current
water content and the maximum holding capacity of the zone. Any additional throughfall enters
storage zone 1 and finally storage zone 2. When all the soil water storage zones are full to capacity,
61
additional precipitation are lost from the system as runoff. Throughfall is assumed to fall evenly
over the main time-step, and is added within the mini-steps of water movement.
2.6.3 Transpiration
Water demand for transpiration is dependent on the soil moisture content of the rooting zone - if a
deficit has developed, then stomatal closure will be initiated (see above). Uptake is also calculated
on a mini-step basis.
Transpiration is reduced if foliage water is present, which must be lost through evaporation before
transpiration can occur. The function weights transpiration of a layer by the proportion of time that
the layer is ‘dry’.
E t = min( Lw , Pe ) + max(0, LPwe ) Φ
2.26
Where Et is evapo-transpiration, Lw, leaf-water, Pe, potential evaporation of a wet surface and Φ, the
transpiration from a dry surface. Transpiration is also governed by boundary layer conductance (ga),
which is dependent on humidity, temperature, and wind-speed. Aerodynamic and boundary layer
conductances are modelled as in MAESTRO (Wang and Jarvis, 1990; Jarvis et al., 1976).
2.7 Respiration
Since the assimilation module generally operates at an hourly timestep, calculations of respiration
are made within this module. Maintenance respiration of each of the main living tissue
compartments (foliage, branch-wood, stem sapwood, transport roots and fine roots) is calculated
using equation 2.27 and expressed on an annual basis as a proportion of tissue dry weight (i.e. kg C
respired per kg carbon in living tissue).
Rt = Rb exp Q10 (t −tb )
2.27
where Rt is the respiration rate at temperature, t, Q10 is the respiratory quotient (default 2.0), and Rb
is the maintenance respiration rate at reference temperature tb.
2.8 Calculation of the Growing season
For deciduous trees, the growing season is defined as the period during which there is green foliage
present on the tree. There is well-documented evidence of growth, elongation, thickening etc of
various organs of the tree at various times of the year (e.g. McWilliam, 1972; Mitrukov, 1976;
Pietarinen et al., 1982). In ForestFlux, these phases are grouped together as the ‘growing season’.
For evergreen species, the definition of growing season is less precise, although there is a period of
defined bud-burst in spring, and in many cases, a dormant period during winter.
2.8.1 Bud-burst
Several types of models of bud-burst were evaluated, including a simple three parameter thermal time
model (Cannell and Smith, 1983), and sequential, alternate and four-phase models which have been
reviewed by Kramer (1994). The most suitable approach, given the limited data available for
validation, is the synthesis model of Hänninen (1990). This is a four-phase model which has been
extended to include the photosensitivity response of Kramer (1994). Unrealistic outputs have been
produced for future climate scenarios, although these problems may be a result of poor input data.
However, it does indicate that without sufficient parameterisation and validation, extrapolation using
the ‘modified synthesis’ model is unwise. In this case, a modified thermal time model, provided more
realistic output, although the model is less sensitive. Either of the models described below can be used,
but care should be taken with parameterisation; this is particularly important for the more complex
62
synthesis model which can easily be over-parameterised, requiring some of the less sensitive
parameters to be fixed.
Both models have a state of chilling (Schl) and forcing (Sfrc):
S chl =
t
Rchl
S frc =
and
t1
t
R frc
2.28
t2
Bud-burst occurs as a function of Schl and Sfrc which are determined by model specific rates of
chilling (Rchl) and forcing (Rfrc). The thermal-time model used in ForestFlux predicts bud-burst
occurs when Sfrc>/=Schl and is expressed as:
0,
ì
R frc = í
îk (T − Tb ),
T ≤ Tb1 ü
ý,
T > Tb 2
t < t1 ü
ý,
t ≥ t2
ì0,
k=í
î1,
and
Rchl = 1
2.29
The parameters t1 and t2 represent the date of the onset of rest and quiescence (namely November 1 and
January 1), with t, the day of year. The parameters, Tb1 and Tb2 are the base temperatures at which
chilling or forcing occurs. In the original thermal time model, Tb1=Tb2=2oC.
The synthesis four-phase model is more complex. The chilling sum, Schl, uses the rate of chilling with
an additional modifier, daylength (Dl) and is calculated from November 1.
Rchl
ü
ï
ï
Tmin < T ≤ Topt ï
ïï
ý
Topt < T ≤ Tmax
ì
0,
ï
ï
ï T − Tmin ,
ïï T − T
min
= ∂ Dl í opt
T − Tmax
ï
,
ï Topt − Tmax
ï
0,
ï
ïî
T ≤ Tmin
2.30
T ≥ Tmax
Modifiers to the chilling and forcing requirements (∆chl, ∆frc)are also calculated from November 1:
∆ chl
ì
0,
ï
ï
ï ∆ c max
=í
(T − Tlow ),
−
T
T
high
low
ï
ï
∆ c max ,
ï
î
ü
ï
ï
ï
≤ T < Thigh ý
T < Tlow
Tlow
2.31
T ≥ Thigh
Similarly,
∆ frc
ì
0
ï
ï
ïï ∆ f max
=í
(T − Tlow )
−
T
T
high
low
ï
ï
∆ f max
ï
îï
ü
ï
ï
ïï
≤ T < Thigh ý
T < Tlow
Tlow
T ≥ Thigh
The chilling and forcing modifiers are used to calculate a competence function, f (C):
63
2.32
ì
0,
ï
ï
ï 1 − C min − ∆ frc
f (C ) = í
( Rchl − C abs ) + C min + ∆ frc ,
ï C crit − ∆ chl − C abs
ï
1,
ï
î
Rchl < C abs
C abs ≤ Rchl < (C crit
ü
ï
ï
ï
− ∆ chl )ý
2.33
Rchl ≥ (C crit − ∆ chl )
The rate of forcing is calculated from the competence function:
R frc
ì
0,
ïï
= f (C ) í
a
ï
,
ïî1 + exp b (T + c )
ü
T ≤ 0ï
ï
ý
T >0
2.34
In this model, bud-burst occurs when the state of forcing, (Sfrc), exceeds a critical threshold,(Fcrit). For
the modified synthesis model, the parameters required are listed in Table 2.1:
Parameter
Minimum temperature for chilling
Tmin
Topt
Optimal temperature for chilling
Tmax
δ
∆ c max
Maximum temperature for chilling
∆ f max
Maximum increase in growth competence from forcing temperature
Tlow
Thigh
Lower temperature value for effect on growth competence
C min
C crit
C abs
a
Minimum growth competence value
Constant for daylength effect
Maximum decrease in chilling requirement for full growth competence.
Highest temperature value for effect on growth competence
Chilling requirement of rest completion
Absolute chilling requirement
Constant for forcing rate
b
Constant for forcing rate
c
Constant for forcing rate
Fcrit
Critical value of the state of forcing to achieve bud-burst
Table 2. 1 Parameter nomenclature for the bud-burst model based on the modified synthesis model of
Hänninen (1990), with photosensitivity of Kramer (1994). See equations 2.30–2.34.
2.8.2 Seasonal Development
Although the development of foliage is assumed to be instantaneous on the date of bud-burst, the
physiological characteristics are assumed to develop as a function of time. The function Dv returns
a value between 0.1 and 1.0 dependent on the number of days since bud-burst, with the maximum
(1.0) occurring 28 days after bud-burst. The function is used as a scalar for the photosynthetic
parameters of Jmax, Vmax and Rd, together with the maximum stomatal conductance, gsmax. The
occurrence of a spring frost retards Dv by seven days. Where no instances of frost occur, the
function is linear.
64
Dv = min((0.1 +
0 .9
28
d i ), 1.0)
2.35
where di is the cumulative number of days since bud-burst (reduced by a maximum of 7 if a spring
frost occurs).
Values of Jmax and Vmax may be determined by an empirical relationship based on nitrogen content
of the foliage. There is no change in photosynthetic capacity associated with the withdrawal of
nutrients in autumn, although all physiological parameters including those describing respiration
can be input at specific time points.
2.8.3 Determination of the end of the growing season
For coniferous trees, the end of growing season is less marked, and in reality, limited
photosynthesis may occur all year round. However, the amount of assimilation during the ‘winter
months’ is small because of low light intensity and temperature, and short day-length.
For deciduous trees, the end of the growing season is defined by the senescence of foliage. There is
a period that precedes this when nutrients are withdrawn from the foliage (leaf yellowing). No
attempt in the model is made to simulate this ‘nutrient withdrawal’. The errors in the calculation of
the overall annual assimilation by this omission are small, as the low-light and temperature result in
minimal photosynthetic CO2 assimilation during this period. Models which attempt to account for
the timing of senescence and the translocation of nutrients usually use a thermal chill function,
which is sometimes coupled with a photo-period function (Pietarinen et al., 1982; Koski and
Sievänen, 1985).
ForestFlux defines leaf longevity as a function of carbon balance (between respiration and
photosynthesis). The modelling of leaf longevity as a function of carbon gain was proposed by
Charbot and Hicks (1982), who expressed it as the difference between photosynthetic rate during
the favourable period of the year and maintenance costs during the unfavourable period, together
with construction costs, and other costs including defence, transport and storage. This approach was
further refined by Kikuzwa (1991, 1995). The assimilation module of ForestFlux contains no
information on tree structure, and thus whole tree carbon balance cannot be used to assess the
longevity of the growing season. An alternative approach has been adopted in which the carbon
balance of the outer-most layer of foliage is assessed, and is defined as:
14
i =1
Ai
≥λ
Ri
2.36
Where Ai and are Ri are the daily carbon assimilation totals and respiration costs of the outermost
layer of foliage in the canopy, and λ is a threshold value. A 14-day moving average is used to
smooth anomalies and prevent the premature prediction of leaf senescence. The end of the growing
season is defined as the day when this 14 day moving average falls below λ. Low temperatures
over-ride this senescence function, with the occurrence of n5 days (T>5oC) or n4 days (T<4oC) also
defining the end of the growing season. The threshold value, λ , also responds to frost immediately
after budburst.
2.9 Initialisation
Most of the input requirements for the assimilation model described above are entered via a single
input file. These inputs use the FORTRAN feature of ‘namelist’. The ‘namelist’ feature allows the
order of inputs (within any list) to be unimportant. The structure of the various lists groups relevant
inputs together; for example, stomatal conductance, soil conditions etc. The input file is in common
with the files used for the growth model, although the growth model parameters need not be set.
65
3.0 Model Validation
Individual modules have been validated using data from a variety of sources. The phenology submodule was calibrated using data from the IPG (International Phenology Garden) project from
1970-1981, for which there was a site at Headley Park in Hampshire, 2 km from the Alice Holt
forest plot. The meteorological data input is for the long-term Alice Holt climatological station,
whilst the budburst data are for Ashtead in Middlesex, UK, courtesy of the Institute of Terrestrial
Ecology, Abbots Wood Research station (Fig. 1a). The correlation between observed and modelled
data is good, although there is a nine day offset which is likely to represent the difference in
temperature between the two sites, which are 50 km apart. In addition, Ashstead is within the
conurbation of Greater London, and consequently subject to an urban heat sink effect.
130
120
110
100
80
1955
modelled
Ashtead
Alice Holt
1965
130
120
110
100
90
80
1975
1985
1995
2005
80
100
120
140
160
modelled budburst (day of year)
year
Figure 1. a) Variation in modelled and observed date of budburst at Alice Holt (1972-1981) and Ashtead
(1960-1996). b) linear regression of observed on modelled date of budburst for Ashtead Surrey. (Y=0.85x +
9.2; r2=0.68). Data for Ashtead courtesy of T. Sparks, Institute of Terrestrial Ecology, Huntingdon, UK).
The carbon and water exchange modules (ie photosynthesis and transpiration) have been validated
using eddy correlation flux data (Figs. 2 and 3). The data shown are for June and July 1998, and
represent fluxes measured at 30 minute intervals. The relationship between measured and modelled
CO2 flux is very good, with a slope of close to unity. However, the modelled flux saturates at
approximately 23 µmol m-2 s-1, which is lower than the maximum observed flux. This may represent
error in the flux measurement system, or a limitation of the model in terms of canopy structure. The
relationship between measured and modelled water flux is not as good as that for CO2. An offset of
approximately 1 mmol m-2 s-1 on the modelled axis is apparent and may indicate the inability of the
eddy correlation system to evaluate water fluxes at low saturation deficits, particularly from a wet
canopy. Interception may therefore be underestimated from the measured flux data. However, in
general terms, the magnitude of the transpiration fluxes agrees well with modelled estimates.
40
flux (µ mol m -2 s -1)
30
20
10
2
measured CO
day of budburst
140
90
140
recorded budburst (day of year)
150
0
-10
-5
0
5
10
15
20
25
30
-10
-20
m odelled CO2 flux (µ
µ m ol m -2 s -1)
Figure 2. Comparison of measured and modelled CO2 fluxes at Alice Holt during June and July 1998
(y=0.989x-1.25; r2=0.85).
66
9
measured H 2O flux (mmol m -2 s -1)
8
7
6
5
4
3
2
1
0
0
-1
1
2
3
4
5
6
7
m odelled H2O flux (m m ol m -2 s -1)
Figure 3. Comparison of measured and modelled water vapour fluxes for June and July 1998 at Alice Holt.
y=0.927x-0.38; r2=0.81.
Further validation of water and CO2 fluxes has been carried out for Norway spruce in Sweden and
beech in Italy (ECOCRAFT, 1999), whilst long-term model simulations have been carried out
through coupling to the growth component of the model, including long term growth of the oak
stand at Alice Holt, Sitka spruce in southern Scotland (Broadmeadow, 2000) and beech in Italy.
References
Broadmeadow M.S.J (2000). Climate Change – Implications for UK Forestry. Forestry Commission
Information Note 31. Forestry Commission, Edinburgh.
Broadmeadow, M.S.J., Heath, J. and Randle, T.J., (1999). Environmental limitations to O3 uptake –
some key results from young trees growing at elevated CO2 concentrations. Water, Air, and
Soil Pollution., 116:299-310.
Cannell, M.G.R. and Smith, R.I., (1983). Thermal time, chill days and prediction of budburst in
Picea sitchensis. Journal of applied Ecology., 20:951-963.
Collares-Pereira, M. and Rabl, A., (1979). The average distribution of solar radiation - correlations
between diffuse and hemispherical and between daily and hourly insolation values. Solar
Energy., 22:155-164.
Chabot,B.F. and Hicks, D.J., (1982). The ecology of leaf life spans. Annual Review of Ecological
Systems., 11:233-260.
De Jong, J.B.R.M., (1980). Een karakterisering van de zonnestraling in Nederland. Doctoraalverslag
Vakgroep Fysische Aspecten van de Gebouwde Omgeving afd. Bouwkunde en Vakgroep
Warmte- en Stromingstechnieken afd. Werktuigbouwkunde, Technische Hogeschool
(Techn. Univ.), Eindhoven, Netherlands, 97+67pp.
ECOCRAFT (1999). Predicted impacts of rising carbon dioxide and temperature on forests in Europe
at stand scale. Final project report (ENV4-CT95-0077).
Erbs, D.G. Kelin, S.A. and Duffie, J.A. (1982). Estimation of the diffuse radiation fraction for
hourly, daily and monthly-average global radiation. Solar Energy., 28:293-302.
Farquhar, G.D. and von Caemmerer, S., (1982). Modelling of photosynthetic response to
environmental conditions. In, Physiological Plant Ecology II. Water Relations and Carbon
67
Assimilation, 12B:550-587; eds, O.L. Lange, C.B. Osmond and H. Ziegler. SpringerVerlag, Berlin.
Hänninen, H., (1990). Modelling bud dormancy release in trees from cool and temperate regions.
Acta Forestalia Fennica., 213:47pp.
Jarvis, P.G., (1976) The interpretation of leaf water potential and stomatal conductance found in
canopies in the field. Philosophical Transactions of the Royal Society of London, Series B.
273:593-690
Jarvis, P.G., James, G.B. and Landsberg, J.J., (1976). Case Study 6: Coniferous Forest. In,
Vegetation and the Atmosphere (Volume 2), 170-240; ed, J.L. Montieth. Academic Press,
London.
Kikuzwa, K., (1991). A cost-benefit analysis of leaf habit and leaf longevity of trees and their
geographical pattern. American Naturalist., 138:1250-1263.
Kikuzwa, K., (1995). Leaf phenology as an optimal strategy for carbon gain in plants. Canadian
Journal of Botany., 73:158-163.
Koski, V. and Sievänen, R., (1985). Timing of growth cessation in relation to the variation of the
growing season. In, Crop Physiology of Forest Trees, 167-193; eds, P.M.A. Tigerstedt, P.
Puttonen and V. Koski. University of Helsinki, University Press.
Kramer, K., (1994). Selecting a model to predict the onset of growth of Fagus sylvatica. Journal of
Applied Ecology. 31:172-181.
Lucas, J.J., (1974). DARE-P, a portable digital simulation. M.Sc. Thesis (Computer Science),
University of Arizona.
Ludlow, A.R., Randle, T.J. and Grace, J.C., (1990). Developing a processed-based growth model
for Sitka spruce. In, Process modelling of forest growth responses to environmental stress,
249-262; eds, R. Dixon, R. Meldahl, G. Ruark and W. Warren. Timber Press, Portland,
Oregon, USA.
McMurtrie, R.E., Rook, D.A. and Kelliher, F.M., (1990). Modelling the yield of Pinus radiata on a
site limited by water and nitrogen. Forest Ecology and Management., 30:381-413.
McWilliam, A., (1972). Some effects of the environment on the growth and development of Picea
sitchensis. PhD thesis, University of Aberdeen.
Medlyn, B.E., Broadmeadow, M.S.J., Randle, T.J., Matteucci, G., and Dufrêne, E., (1999). Model
Comparison. In, Predicted Impacts of Rising Carbon Dioxide and Temperature on Forests
in Europe at Stand Scale, 97-105; Ecocraft final report ENV4-CT95-007 DG. XII, Brussels.
Meng, F.R. and Arp, P.A., (1994). Modelling photosynthetic responses of a spruce canopy to SO2
exposure. Forest Ecology and Management., 67:69-85.
Mitrukov, A.E., 1976. Seasonal growth of shoots, needles and radial increment of the stem, in
stands of the Pinetum cladinosum of S. Karelia. Lesovedenie., 3:44-49.
Neumann, H.H., Hartog, G.D. and Shaw, R.H., (1989). Leaf area measurements based on
hemispheric photographs and leaf-litter collection in a deciduous forest during autumn leaffall. Agricultural and Forest Meteorology., 45:325-335.
68
Pietarinen, I., Kanninen, M., Hari, P. and Kellomaki, S., (1982). A simulation model for daily
growth of shoots, needles, and stem diameter in Scots pine trees. Forest Science., 28:573581.
Randle, T.J., (1997). Obtaining hourly radiation from daily sunshine-hour records. Internal working
paper, Forest Research.
Randle, T.J. and Ludlow, A.R., (1998). Growth costs to plantations: modelling the cumulative effect
of defoliation by Elatobium abietinum. In, The Green Spruce Aphid in Western Europe:
Ecology, Status, Impacts and Prospects for Management, 32-40; eds, K. Day, G.
Halldorsson, S. Harding and N. Straw. Forestry Commission Technical Paper 24, Forestry
Commission, Edinburgh.
Shinozaki, K., Yoda, K., Hozumi, K. and Kira, T., (1964). A quantitative analysis of plant form –
the pipe model theory. I. Basic analysis. Japanese Journal of Ecology., 14:97-105.
Spitters, C.J.T., (1986a). Separating the diffuse and direct component of global radiation and its
implications for modelling canopy photosynthesis. Part I. Components of incoming
radiation. Agricultural and Forest Meteorology., 38:217-229.
Spitters, C.J.T., (1986b). Separating the diffuse and direct component of global radiation and its
implications for modelling canopy photosynthesis. Part II. Calculation of canopy
photosynthesis. Agricultural and Forest Meteorology., 38:231-242.
von Caemmerer, S., Evans, J.R., Hudson, G.S., and Andrews, T.J., (1994). The kinetics of ribulose1,5-biphosphate carboxylase/oxygenase in vivo inferred from measurements of
photosynthesis in leaves of transgenic tobacco. Planta., 195:88-97.
Wait, J.V. and Clarke, D., (1978). DARE P Users’ manual. (Version 4.2). Department of Electrical
Engineering, University of Arizona.
Wang, Y-P. and Jarvis, P.G., (1990) Description and validation of an array model – MAESTRO.
Agricultural and Forest Meteorology., 51:257-280.
Weiss, A. and Norman, J.M., (1985). Partitioning solar radiation into direct and diffuse, visible and
near-infrared components. Agricultural and Forest Meteorology.,34:205-213.
69
Appendix II – Spring phenology reporting forms
SPRING PHENOLOGY IN LEVEL II PLOTS (FLUSHING)
•
•
record operator details, plot and date
observe tree from aspect given in table below, and only from this position
TREE NUMBER
103
111
120
123
124
201
202
203
204
DIRECTION OF ASSESSMENT
•
record the progress of flushing of each tree, as the percentage of the crown that is
showing any development according to the scoring system given - please note that no
development scores 1
• record the average stage of development of each tree according to the scoring system
given - green bits, identifiable leaf forms, or leaves fully expanded
• record any damage, particularly from frost on form 11a
• please add any additional comments on the bottom of form 11a, making sure to include
date, and relevant tree number
70
LEVEL II SPRING (FLUSHING) PHENOLOGY. FORM 11b
SURVEYOR NAME/CODE:_____________________________
SPECIES CODE:___________
TABLE A:
tree number
132
139
177
208
209
401
402
403
404
405
406
SPECIES NAME:__________________________
PLOT NAME:________________
DATE (MM/YY):_____________
PROGRESS OF FLUSHING
1
TABLE B:
tree number
132
139
177
208
209
401
402
403
404
405
406
PLOT CODE:________
2
3
4
5
6
7
8
9
10
11
12
13
14
Day of month
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
7
8
9
10
11
12
13
14
Day of month
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
DEVELOPMENTAL STAGE
1
2
3
4
5
6
TABLE A
TABLE B
1
2
3
4
5
5
6
7
ABSENCE OF FLUSHING
PRESENT BUT INFREQUENT (LESS THAN 33% OF CROWN)
PRESENT AND COMMON (33-66% OF CROWN)
PRESENT AND ABUNDANT (MORE THAN 66% OF CROWN)
COMPLETED (COMPLETE CROWN FLUSHING)
71
GREEN BITS SHOWING
DEFINITE LEAVES VISIBLE
LEAVES FULLY EXPANDED
LEVEL II SPRING (FLUSHING) PHENOLOGY. FORM 11b*
SURVEYOR NAME/CODE:_____________________________
SPECIES CODE:___________
TABLE A:
tree number
132
139
177
208
209
401
402
403
404
405
406
SPECIES NAME:__________________________
PLOT NAME:________________
DATE (MM/YY):_____________
PROGRESS OF FLUSHING
1
TABLE B:
tree number
132
139
177
208
209
401
402
403
404
405
406
PLOT CODE:________
2
3
4
5
6
7
8
9
10
11
12
13
14
Day of month
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
7
8
9
10
11
12
13
14
Day of month
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
DEVELOPMENTAL STAGE
1
2
3
4
5
6
TABLE A
TABLE B
1
2
3
4
5
1
2
3
4
5
ABSENCE OF FLUSHING
PRESENT BUT INFREQUENT (LESS THAN 33% OF CROWN)
PRESENT AND COMMON (33-66% OF CROWN)
PRESENT AND ABUNDANT (MORE THAN 66% OF CROWN)
COMPLETED (COMPLETE CROWN FLUSHING)
72
BUDS DORMANT
BUDS EXPANDED (AND PROBABLY BROKEN)
GREEN BITS SHOWING
DEFINITE LEAVES VISIBLE
LEAVES FULLY EXPANDED
Appendix III: Pictorial representation of standard flushing stages
73
Photographs of the three primary stages of flushing in oak: top ‘green bits showing’; middle ‘definite
leaves visible’; bottom ‘leaves fully expanded. See Table 11b.
74
Appendix IV Glossary of abbreviations and definition of symbols
AWS
DBH
ECN
GPP
GROMIT
GYC
LAI
NPP
PAR
RH
RuBisCO
SLA
SVP
TDR
VOC
Automatic weather station
Diameter at breast height
Environmental Change Network
Gross Primary productivity
Growth model of individual trees
General yield class
Leaf area index (leaf area per unit ground area)
Net primary productivity
Photosynthetically active radiation
Relative humidity
Ribulose bis-phosphate carboxylase-oxygenase
Specific leaf area (leaf area per unit weight)
Saturated vapour pressure
Time domain reflectometry
Volatile organic compound
Ei
Et
gsmax
Ileaf
Jmax
Kext
Lmax
[N]
[N]a
Q
Qinc
Qtrans
R
Rd
Rf
Rl
Rw
SMmin
SMsat
τ
Ta
Tmax
Tmin
Ts
Vmax
vp
ws
evaporative component of evapo-transpiration
transpiratory component of evapo-transpiration
stomatal conductance in the absence of limiting factors
leaf water holding capacity
potential electron transport rate
effective light extinction coefficient
maximum leaf area index
foliar nitrogen content expressed on a dry weight basis
foliar nitrogen content expressed on an area basis
solar radiation
incident solar radiation
transmitted solar radiation
rainfall
foliar dark respiration in the presence of light
foliar respiration
correction factor for leaf decomposition contribution to Rw
stand wood respiration (including stem, branches and roots)
minimum soil moisture content for stomatal conductance
saturated soil moisture content (field capacity)
canopy light transmission
air temperature
maximum daily air temperature
minimum daily air temperature
soil temperature
maximum RuBisCO activity
vapour pressure
wind speed
75